Using Time Series Predictive Models for Early Detection of Gambling Addiction in Problem Gamblers

Data Gathering

import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

daily_agg_df = pd.read_csv('./datasets/Raw Datset II.Daily aggregates_Gray_LaPlante_PAB_2012.dat', delimiter='\t')
rg_det_df = pd.read_csv('./datasets/Raw Datset III.Responsible gambling details_Gray_LaPlante_PAB_2012.dat', delimiter='\t')
demog_df = pd.read_csv('./datasets/Raw Datset I.Demographics_Gray_LaPlante_PAB_2012.dat', delimiter='\t')

C:\Users\abhiv\AppData\Local\Temp\ipykernel_14916\3515817822.py:5: DtypeWarning: Columns (3,4,5) have mixed types. Specify dtype option on import or set low_memory=False.
  daily_agg_df = pd.read_csv('./datasets/Raw Datset II.Daily aggregates_Gray_LaPlante_PAB_2012.dat', delimiter='\t')

rg_det_df

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

</style>

	UserID	RGsumevents	RGFirst_Date	RGLast_date	Event_type_first	Interventiontype_first
0	2169867	1	11/19/2009	11/19/2009	9	18
1	7035862	1	11/15/2009	11/15/2009	9	18
2	5911218	1	11/8/2009	11/8/2009	9	18
3	5872708	1	11/3/2009	11/3/2009	9	18
4	5746942	2	10/18/2009	11/3/2009	9	18
...	...	...	...	...	...	...
2063	2451840	5		11/3/2009	12
2064	9140426	1	8/19/2009	8/19/2009	10
2065	2590026	2	8/5/2009	8/6/2009	10
2066	1023918	1	4/9/2009	4/9/2009	6
2067	6691324	1	12/19/2008	12/19/2008	1

2068 rows × 6 columns

demog_df

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

</style>

	USERID	RG_case	CountryName	LanguageName	Gender	YearofBirth	Registration_date	First_Deposit_Date
0	2975944	1	Germany	German	M	1970	6/2/2006	6/9/2006
1	9822065	1	Germany.COM	German	F	1963	11/21/2009	11/21/2009
2	9622454	1	France.COM	French	F	1981	10/19/2009	10/19/2009
3	9619356	1	Italy.IT	Italian	F	1975	10/18/2009	10/18/2009
4	9593498	1	Germany.COM	German	F	1990	10/14/2009	10/14/2009
...	...	...	...	...	...	...	...	...
4129	107292	0	Austria	German	M	1975	7/9/2000	7/9/2000
4130	92140	0	Austria	German	M	1973	6/25/2000	6/25/2000
4131	80281	0	Austria	German	M	1970	6/13/2000	6/13/2000
4132	74438	0	Austria	German	M	1975	6/9/2000	6/9/2000
4133	36822	0	Austria	German	M	1970	3/20/2000	5/8/2000

4134 rows × 8 columns

daily_agg_df

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

</style>

	UserID	Date	ProductType	Turnover	Hold	NumberofBets
0	31965	5/8/2000	1	15.3388	15.3388	1
1	31965	5/10/2000	1	34.1594	34.1594	5
2	31965	5/18/2000	1	24.5419	24.5419	4
3	31965	5/22/2000	1	2.5309	2.5309	1
4	31965	5/23/2000	1	15.3387	15.3387	2
...	...	...	...	...	...	...
981777	9200696	10/12/2010	25			95
981778	7912483	9/23/2010	25			60
981779	9200696	10/11/2010	25			2
981780	9200696	10/14/2010	25			2
981781	9200696	10/24/2010	25			2

981782 rows × 6 columns

Data Transformation and Cleaning

import pandas as pd

# Define a standard date for filling empty and invalid cells
standard_date = pd.to_datetime('01/01/1900', format='%d/%m/%Y', errors='coerce')

# Fill empty and invalid cells with the standard date
daily_agg_df['Date'] = pd.to_datetime(daily_agg_df['Date'], errors='coerce').fillna(standard_date)
demog_df['Registration_date'] = pd.to_datetime(demog_df['Registration_date'], errors='coerce').fillna(standard_date)
demog_df['First_Deposit_Date'] = pd.to_datetime(demog_df['First_Deposit_Date'], errors='coerce').fillna(standard_date)
rg_det_df['RGFirst_Date'] = pd.to_datetime(rg_det_df['RGFirst_Date'], errors='coerce').fillna(standard_date)
rg_det_df['RGLast_date'] = pd.to_datetime(rg_det_df['RGLast_date'], errors='coerce').fillna(standard_date)

# Create new datetime columns
daily_agg_df['Aggregate_Date'] = pd.to_datetime(daily_agg_df['Date'])

daily_agg_df.drop('Date', axis=1, inplace=True)

demog_df['Registration_date'] = pd.to_datetime(demog_df['Registration_date'])
demog_df['First_Deposit_Date'] = pd.to_datetime(demog_df['First_Deposit_Date'])
rg_det_df['RGFirst_Date'] = pd.to_datetime(rg_det_df['RGFirst_Date'])
rg_det_df['RGLast_date'] = pd.to_datetime(rg_det_df['RGLast_date'])

# Rename the 'old_column_name' to 'new_column_name'
daily_agg_df = daily_agg_df.rename(columns={'UserID': 'UserID'})
demog_df = demog_df.rename(columns={'USERID': 'UserID'})
rg_det_df = rg_det_df.rename(columns={'UserID': 'UserID'})

daily_agg_df
product_type_frequencies = daily_agg_df['ProductType'].value_counts()
print(product_type_frequencies)

1     399410
2     331828
10    127223
8      37749
15     25646
4      20749
6      13558
3       7539
14      7310
19      6122
7       1741
23      1215
5        559
17       506
20       321
22       158
9         67
21        38
24        35
25         7
16         1
Name: ProductType, dtype: int64

daily_agg_df_t=daily_agg_df.tail(10)

daily_agg_df_t

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

</style>

	UserID	ProductType	NumberofBets	Aggregate_Date
981772	4608302	24	6	2010-09-17
981773	1285995	24	2	2010-07-27
981774	4608302	24	3	2010-09-16
981775	7912483	25	393	2010-09-24
981776	7912483	25	228	2010-09-22
981777	9200696	25	95	2010-10-12
981778	7912483	25	60	2010-09-23
981779	9200696	25	2	2010-10-11
981780	9200696	25	2	2010-10-14
981781	9200696	25	2	2010-10-24

demog_df

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

</style>

	UserID	RG_case	CountryName	LanguageName	Gender	YearofBirth	Registration_date	First_Deposit_Date
0	2975944	1	Germany	German	M	1970	2006-06-02	2006-06-09
1	9822065	1	Germany.COM	German	F	1963	2009-11-21	2009-11-21
2	9622454	1	France.COM	French	F	1981	2009-10-19	2009-10-19
3	9619356	1	Italy.IT	Italian	F	1975	2009-10-18	2009-10-18
4	9593498	1	Germany.COM	German	F	1990	2009-10-14	2009-10-14
...	...	...	...	...	...	...	...	...
4129	107292	0	Austria	German	M	1975	2000-07-09	2000-07-09
4130	92140	0	Austria	German	M	1973	2000-06-25	2000-06-25
4131	80281	0	Austria	German	M	1970	2000-06-13	2000-06-13
4132	74438	0	Austria	German	M	1975	2000-06-09	2000-06-09
4133	36822	0	Austria	German	M	1970	2000-03-20	2000-05-08

4134 rows × 8 columns

rg_det_df

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

</style>

	UserID	RGsumevents	RGFirst_Date	RGLast_date	Event_type_first	Interventiontype_first
0	2169867	1	2009-11-19	2009-11-19	9	18
1	7035862	1	2009-11-15	2009-11-15	9	18
2	5911218	1	2009-11-08	2009-11-08	9	18
3	5872708	1	2009-11-03	2009-11-03	9	18
4	5746942	2	2009-10-18	2009-11-03	9	18
...	...	...	...	...	...	...
2063	2451840	5	1900-01-01	2009-11-03	12
2064	9140426	1	2009-08-19	2009-08-19	10
2065	2590026	2	2009-08-05	2009-08-06	10
2066	1023918	1	2009-04-09	2009-04-09	6
2067	6691324	1	2008-12-19	2008-12-19	1

2068 rows × 6 columns

from sklearn.preprocessing import LabelEncoder

# Initialize label encoders
label_encoder_country = LabelEncoder()
label_encoder_language = LabelEncoder()
label_encoder_gender = LabelEncoder()

# Fit and transform the categorical columns
demog_df['CountryName'] = label_encoder_country.fit_transform(demog_df['CountryName'])
demog_df['LanguageName'] = label_encoder_language.fit_transform(demog_df['LanguageName'])
demog_df['Gender'] = label_encoder_gender.fit_transform(demog_df['Gender'])

demog_df

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

</style>

	UserID	RG_case	CountryName	LanguageName	Gender	YearofBirth	Registration_date	First_Deposit_Date
0	2975944	1	18	8	1	1970	2006-06-02	2006-06-09
1	9822065	1	19	8	0	1963	2009-11-21	2009-11-21
2	9622454	1	17	7	0	1981	2009-10-19	2009-10-19
3	9619356	1	25	11	0	1975	2009-10-18	2009-10-18
4	9593498	1	19	8	0	1990	2009-10-14	2009-10-14
...	...	...	...	...	...	...	...	...
4129	107292	0	4	8	1	1975	2000-07-09	2000-07-09
4130	92140	0	4	8	1	1973	2000-06-25	2000-06-25
4131	80281	0	4	8	1	1970	2000-06-13	2000-06-13
4132	74438	0	4	8	1	1975	2000-06-09	2000-06-09
4133	36822	0	4	8	1	1970	2000-03-20	2000-05-08

4134 rows × 8 columns

Merging of Datasets

merged_df = daily_agg_df.merge(demog_df, on='UserID', how='outer')
merged_df = merged_df.merge(rg_det_df, on='UserID', how='outer')

merged_df['RG_case'].value_counts()

1    811570
0    170233
Name: RG_case, dtype: int64

merged_df

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

</style>

	UserID	ProductType	Turnover	Hold	NumberofBets	Aggregate_Date	RG_case	CountryName	LanguageName	Gender	YearofBirth	Registration_date	First_Deposit_Date	RGsumevents	RGFirst_Date	RGLast_date	Event_type_first	Interventiontype_first
0	31965	1.0	15.3388	15.3388	1	2000-05-08	1	19	8	1	1971	1999-09-17	2000-05-08	1.0	2009-03-04	2009-03-04	2.0	8
1	31965	1.0	34.1594	34.1594	5	2000-05-10	1	19	8	1	1971	1999-09-17	2000-05-08	1.0	2009-03-04	2009-03-04	2.0	8
2	31965	1.0	24.5419	24.5419	4	2000-05-18	1	19	8	1	1971	1999-09-17	2000-05-08	1.0	2009-03-04	2009-03-04	2.0	8
3	31965	1.0	2.5309	2.5309	1	2000-05-22	1	19	8	1	1971	1999-09-17	2000-05-08	1.0	2009-03-04	2009-03-04	2.0	8
4	31965	1.0	15.3387	15.3387	2	2000-05-23	1	19	8	1	1971	1999-09-17	2000-05-08	1.0	2009-03-04	2009-03-04	2.0	8
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
981798	1190813	NaN	NaN	NaN	NaN	NaT	0	0	0	1		1900-01-01	2005-07-01	NaN	NaT	NaT	NaN	NaN
981799	1622440	NaN	NaN	NaN	NaN	NaT	0	0	0	1		1900-01-01	2005-05-21	NaN	NaT	NaT	NaN	NaN
981800	1108530	NaN	NaN	NaN	NaN	NaT	0	0	0	1		1900-01-01	2004-08-22	NaN	NaT	NaT	NaN	NaN
981801	683142	NaN	NaN	NaN	NaN	NaT	0	0	0	1		1900-01-01	2003-05-20	NaN	NaT	NaT	NaN	NaN
981802	113041	NaN	NaN	NaN	NaN	NaT	0	0	0	1		1900-01-01	2000-07-29	NaN	NaT	NaT	NaN	NaN

981803 rows × 18 columns

merged_df_tsc=merged_df.drop(columns=['RGLast_date','RGFirst_Date','Registration_date','First_Deposit_Date'])
merged_df_tsc=merged_df_tsc #.tail(13000)
filtered_df = merged_df_tsc[merged_df_tsc['ProductType'] == 2]
filtered_df
# Sort the DataFrame by 'user_id' and 'date'
filtered_df = filtered_df.sort_values(by=['UserID', 'Aggregate_Date'])

filtered_df['RG_case'].value_counts()

1    294375
0     37453
Name: RG_case, dtype: int64

subset_columns = ['Aggregate_Date', 'UserID']

# Identify and drop duplicate rows based on the specified subset of columns
filtered_df = filtered_df.drop_duplicates(subset=subset_columns, keep='first')

# Display the DataFrame after dropping duplicates
print("DataFrame after dropping duplicates:")
print(filtered_df)

DataFrame after dropping duplicates:
         UserID  ProductType Turnover   Hold NumberofBets Aggregate_Date  \
402       31965          2.0     20.0   20.0            1     2002-11-12   
1002      31965          2.0    73.18  18.96            6     2002-11-14   
403       31965          2.0     10.0   10.0            1     2002-11-15   
1217      31965          2.0   163.28   19.0            9     2002-11-16   
1331      31965          2.0   162.34  -97.5           13     2002-11-17   
...         ...          ...      ...    ...          ...            ...   
973451  9822065          2.0     10.0   10.0            1     2010-02-09   
973463  9822065          2.0     15.0  -3.35            1     2010-03-02   
973452  9822065          2.0      1.0    1.0            1     2010-04-21   
973456  9822065          2.0      1.0   -2.8            1     2010-06-10   
973516  9859152          2.0     13.0   13.0            5     2009-11-27   

        RG_case  CountryName  LanguageName  Gender YearofBirth  RGsumevents  \
402           1           19             8       1        1971          1.0   
1002          1           19             8       1        1971          1.0   
403           1           19             8       1        1971          1.0   
1217          1           19             8       1        1971          1.0   
1331          1           19             8       1        1971          1.0   
...         ...          ...           ...     ...         ...          ...   
973451        1           19             8       0        1963          1.0   
973463        1           19             8       0        1963          1.0   
973452        1           19             8       0        1963          1.0   
973456        1           19             8       0        1963          1.0   
973516        0           19             8       1        1982          NaN   

        Event_type_first Interventiontype_first  
402                  2.0                      8  
1002                 2.0                      8  
403                  2.0                      8  
1217                 2.0                      8  
1331                 2.0                      8  
...                  ...                    ...  
973451               4.0                     13  
973463               4.0                     13  
973452               4.0                     13  
973456               4.0                     13  
973516               NaN                    NaN  

[322898 rows x 14 columns]

filtered_df['RG_case'].value_counts()

null_mask = filtered_df.isna()

# Use sum() to count the null values in each column
null_count = null_mask.sum()

# Display columns with null values and their respective counts
print("Columns with null values and their counts:")
print(null_count[null_count > 0])
filtered_df

filtered_df = filtered_df.fillna(0)

Columns with null values and their counts:
RGsumevents               36498
Event_type_first          36498
Interventiontype_first    36498
dtype: int64

filtered_df
# Convert birth year to age until 2010
current_year = 2010
filtered_df['Age_until_2010'] = filtered_df['YearofBirth'].apply(lambda birth_year: current_year - int(birth_year))

Feature Selection

columns_for_heatmap = filtered_df.columns.difference(['ProductType'])

# Create a subset DataFrame with selected columns
heatmap_data = filtered_df[columns_for_heatmap]

# Create a heatmap using seaborn
plt.figure(figsize=(12, 8))
sns.heatmap(heatmap_data.corr(), annot=True, cmap='coolwarm', fmt=".2f")
plt.title('Heatmap of Correlation Matrix')
plt.show()

C:\Users\abhiv\AppData\Local\Temp\ipykernel_14916\2072689473.py:8: FutureWarning: The default value of numeric_only in DataFrame.corr is deprecated. In a future version, it will default to False. Select only valid columns or specify the value of numeric_only to silence this warning.
  sns.heatmap(heatmap_data.corr(), annot=True, cmap='coolwarm', fmt=".2f")

filtered_df

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

</style>

	UserID	ProductType	Turnover	Hold	NumberofBets	Aggregate_Date	RG_case	CountryName	LanguageName	Gender	YearofBirth	RGsumevents	Event_type_first	Interventiontype_first	Age_until_2010
402	31965	2.0	20.0	20.0	1	2002-11-12	1	19	8	1	1971	1.0	2.0	8	39
1002	31965	2.0	73.18	18.96	6	2002-11-14	1	19	8	1	1971	1.0	2.0	8	39
403	31965	2.0	10.0	10.0	1	2002-11-15	1	19	8	1	1971	1.0	2.0	8	39
1217	31965	2.0	163.28	19.0	9	2002-11-16	1	19	8	1	1971	1.0	2.0	8	39
1331	31965	2.0	162.34	-97.5	13	2002-11-17	1	19	8	1	1971	1.0	2.0	8	39
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
973451	9822065	2.0	10.0	10.0	1	2010-02-09	1	19	8	0	1963	1.0	4.0	13	47
973463	9822065	2.0	15.0	-3.35	1	2010-03-02	1	19	8	0	1963	1.0	4.0	13	47
973452	9822065	2.0	1.0	1.0	1	2010-04-21	1	19	8	0	1963	1.0	4.0	13	47
973456	9822065	2.0	1.0	-2.8	1	2010-06-10	1	19	8	0	1963	1.0	4.0	13	47
973516	9859152	2.0	13.0	13.0	5	2009-11-27	0	19	8	1	1982	0.0	0.0	0	28

322898 rows × 15 columns

Model Fitting

K - Means Clustering

import numpy as np
import pandas as pd
from sklearn.cluster import KMeans
from sklearn.decomposition import PCA
import matplotlib.pyplot as plt


# Pivot the DataFrame to create a 3D array with entries as rows, features as columns, and dates as depth
user_data_3d = filtered_df.pivot(index='UserID', columns='Aggregate_Date', values=['Turnover', 'Hold', 'NumberofBets','Age_until_2010', 'Interventiontype_first','CountryName','Gender','Event_type_first'])

# Fill missing values with zeros (if any)
user_data_3d = user_data_3d.fillna(0)

# Convert empty strings (' ') to float values of zero (0)
user_data_3d = user_data_3d.replace(' ', 0).astype(float)
user_data_3d = user_data_3d.astype(float)

# Convert the pivot table to a NumPy array
user_data_array = user_data_3d.to_numpy()

#k fold value
k = 3

# Perform K-means clustering
model = KMeans(n_clusters=k, random_state=0)
y_pred = model.fit_predict(user_data_array)

# Apply PCA to reduce dimensionality to 2D
pca = PCA(n_components=2)
user_data_pca = pca.fit_transform(user_data_array)

# Visualize the clustered entries using PCA components
plt.figure(figsize=(8, 6))
scatter = plt.scatter(user_data_pca[:, 0], user_data_pca[:, 1], c=y_pred, cmap='viridis')
plt.xlabel('PCA Component 1')
plt.ylabel('PCA Component 2')



    
cluster_dict={}

# Create a dictionary to store user IDs and cluster number
cluster_dict = {i: {'users': [], 'cluster_num': i} for i in range(k)}

# Loop through the cluster labels and append user IDs to the corresponding cluster
for user_id, cluster_label in zip(user_data_3d.index, y_pred):
    cluster_dict[cluster_label]['users'].append(user_id)

# Print the user IDs and cluster number in each cluster
for cluster_label, info in cluster_dict.items():
    print(f'Cluster {cluster_label} (Cluster {info["cluster_num"]}):\nUsers: {info["users"]}')
    
    
# Create a legend for the cluster numbers
legend_labels = ['Moderate Problem Gamblers','Early Players','Problem Gamblers']
legend = plt.legend(handles=scatter.legend_elements()[0], title='Cluster', labels=legend_labels)

plt.gca().add_artist(legend)

plt.show()

C:\Users\abhiv\AppData\Local\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:870: FutureWarning: The default value of `n_init` will change from 10 to 'auto' in 1.4. Set the value of `n_init` explicitly to suppress the warning
  warnings.warn(


Cluster 0 (Cluster 0):
Users: [868583, 1175809, 1411743, 1457496, 1486136, 1662632, 1679490, 1776178, 1790848, 1921204, 2070894, 2150296, 2155065, 2589710, 2704382, 2852203, 3669466, 3852889, 3904422, 3968386, 4006708, 4371320, 4394754, 4412550, 4495603, 4532357, 5106620, 5308271, 5488160, 5660719, 5678852, 5723033, 6158120, 6175402, 6239380, 6283338, 6709379, 6985339, 7192925]
Cluster 1 (Cluster 1):
Users: [31965, 32639, 36822, 36916, 74438, 90746, 91707, 92140, 96950, 99596, 100167, 107124, 107292, 109445, 113136, 119406, 121115, 134759, 157710, 165200, 165335, 166143, 178914, 187808, 191415, 212099, 217121, 221079, 233644, 275466, 306915, 332091, 348381, 349493, 349855, 350228, 355882, 366074, 375779, 376200, 377204, 378707, 380006, 380781, 386047, 399987, 402872, 403944, 405981, 411101, 420681, 441202, 452829, 455766, 464507, 465623, 467890, 468786, 469615, 470323, 471813, 484593, 485942, 486333, 486555, 486829, 502856, 513616, 518169, 519601, 528696, 535025, 537380, 539496, 548298, 551872, 555911, 562883, 572029, 576509, 578303, 578678, 579708, 581568, 596939, 600062, 601929, 606177, 606833, 607958, 610512, 610990, 612580, 614867, 615124, 617512, 617643, 618006, 618134, 622333, 623222, 624243, 626809, 626818, 626970, 627380, 627907, 628673, 636671, 637178, 637789, 637922, 638301, 640126, 640946, 644434, 646063, 649274, 651518, 651781, 652803, 654065, 654455, 656052, 659457, 660104, 662035, 668204, 668658, 669331, 669759, 671003, 671210, 671657, 671821, 676437, 676439, 679909, 689277, 691215, 691219, 691300, 693201, 695820, 704670, 707265, 708122, 709000, 709426, 709566, 712428, 713486, 714120, 715823, 716725, 717922, 719989, 721385, 722705, 724802, 725520, 727358, 731018, 734038, 734064, 735381, 740719, 758982, 759655, 767729, 779314, 779784, 781520, 793809, 809647, 815193, 816423, 819431, 819712, 820287, 821915, 823927, 829363, 832139, 836619, 838424, 839105, 841112, 842884, 845248, 847728, 852596, 859526, 860020, 860767, 863069, 872943, 872995, 876092, 876669, 876762, 876836, 877236, 878190, 879490, 881586, 891132, 893392, 893855, 894982, 897560, 905798, 908460, 910113, 910143, 912480, 912981, 921177, 921652, 934074, 939165, 939794, 940344, 943528, 945804, 946810, 951260, 954936, 954947, 958509, 960009, 962369, 963330, 965698, 969529, 971617, 978164, 984545, 984819, 986088, 986289, 991634, 992202, 994151, 998680, 999018, 1002807, 1003282, 1006572, 1006668, 1015042, 1019680, 1023918, 1029762, 1033839, 1034626, 1038912, 1039083, 1052452, 1053169, 1054028, 1055097, 1057341, 1059479, 1065910, 1066411, 1066905, 1068147, 1068353, 1068512, 1068666, 1069445, 1069480, 1069718, 1073054, 1073059, 1076465, 1077256, 1086185, 1086610, 1086719, 1089300, 1092942, 1097302, 1098879, 1098995, 1099733, 1100329, 1102048, 1102347, 1104183, 1104232, 1105726, 1108323, 1108391, 1108768, 1110796, 1110869, 1112009, 1112125, 1114773, 1119668, 1119951, 1121270, 1121292, 1123515, 1124321, 1124617, 1125602, 1128246, 1128286, 1128557, 1129130, 1131579, 1133624, 1134790, 1137771, 1139267, 1143257, 1145088, 1152572, 1153292, 1161759, 1166481, 1168835, 1169296, 1171028, 1174435, 1177014, 1177425, 1178665, 1179202, 1180532, 1184758, 1186045, 1188402, 1188702, 1188944, 1188981, 1190225, 1190678, 1190696, 1190818, 1190880, 1191041, 1191110, 1191496, 1191525, 1191883, 1195259, 1195708, 1198551, 1198657, 1200490, 1202145, 1205205, 1207102, 1208186, 1210080, 1211034, 1211153, 1215396, 1216806, 1216960, 1217056, 1217347, 1217527, 1217606, 1218000, 1220034, 1221797, 1222876, 1226174, 1226493, 1226683, 1227473, 1227499, 1229500, 1230025, 1230735, 1236874, 1247913, 1251523, 1260516, 1260523, 1261748, 1261993, 1262533, 1263607, 1265820, 1267246, 1268833, 1269295, 1269446, 1269588, 1273875, 1275992, 1276276, 1276696, 1277104, 1277167, 1277939, 1278165, 1280420, 1280671, 1281272, 1282148, 1282506, 1284616, 1285413, 1285959, 1285995, 1286146, 1286507, 1288342, 1289020, 1289575, 1290126, 1290225, 1290981, 1292031, 1292371, 1292648, 1295199, 1298752, 1298980, 1299294, 1299793, 1301234, 1301389, 1304180, 1305631, 1305661, 1306998, 1307000, 1307421, 1307533, 1309623, 1311762, 1315386, 1316055, 1316554, 1317059, 1318469, 1320903, 1321102, 1323207, 1325161, 1325647, 1325682, 1329184, 1330470, 1333978, 1334455, 1336326, 1336704, 1337767, 1339503, 1342111, 1342426, 1343436, 1344751, 1344830, 1345978, 1348134, 1348150, 1348549, 1353206, 1356035, 1357532, 1361415, 1362339, 1362354, 1362573, 1363362, 1363440, 1364127, 1364138, 1365118, 1365304, 1367719, 1371692, 1374073, 1374185, 1375149, 1376027, 1376313, 1376503, 1377425, 1377479, 1381040, 1383267, 1387059, 1387925, 1388917, 1390642, 1393636, 1394702, 1397690, 1398079, 1398948, 1402673, 1402736, 1403471, 1406089, 1406743, 1407020, 1407759, 1411882, 1412153, 1412349, 1415558, 1416109, 1418087, 1420230, 1422024, 1423249, 1425148, 1426789, 1426870, 1427115, 1427840, 1428100, 1428339, 1430272, 1430651, 1431358, 1436194, 1444002, 1444024, 1444638, 1444690, 1446765, 1446979, 1447394, 1448747, 1448755, 1448966, 1448980, 1449058, 1453530, 1455429, 1455607, 1458694, 1460790, 1461123, 1466020, 1467698, 1467720, 1468506, 1474080, 1475521, 1475836, 1476237, 1476815, 1477587, 1481182, 1482265, 1487056, 1487262, 1489176, 1494069, 1495100, 1499093, 1504089, 1504941, 1511327, 1511559, 1511812, 1516266, 1516433, 1516508, 1519947, 1521035, 1521486, 1521811, 1522428, 1522618, 1524410, 1524893, 1527006, 1528877, 1529711, 1531097, 1533580, 1533943, 1538409, 1539163, 1539371, 1540796, 1541252, 1541799, 1545811, 1546185, 1548350, 1548547, 1549759, 1551158, 1556539, 1556993, 1559015, 1560024, 1560193, 1561422, 1564451, 1567292, 1569967, 1572636, 1573032, 1575816, 1578866, 1579804, 1586572, 1587474, 1589646, 1589909, 1590508, 1590556, 1592977, 1594233, 1594288, 1599117, 1600042, 1601073, 1601397, 1602071, 1603442, 1604150, 1606117, 1609008, 1610238, 1610754, 1613141, 1613727, 1614409, 1615173, 1618935, 1624023, 1626239, 1630307, 1630590, 1645647, 1646641, 1648376, 1649239, 1651030, 1651912, 1652164, 1659620, 1663129, 1664927, 1666227, 1669102, 1669496, 1671815, 1677908, 1680006, 1680615, 1683379, 1687563, 1689908, 1690828, 1691317, 1692544, 1693518, 1696904, 1699984, 1700174, 1711860, 1713213, 1715092, 1717071, 1717825, 1720080, 1723411, 1725647, 1729399, 1730948, 1731044, 1735247, 1738054, 1738946, 1739322, 1740517, 1740827, 1741128, 1741291, 1742295, 1745411, 1745784, 1747125, 1751937, 1752640, 1753291, 1755673, 1759481, 1759742, 1761005, 1763045, 1764945, 1766119, 1766299, 1766392, 1768666, 1770262, 1775842, 1776268, 1776887, 1776969, 1777482, 1780205, 1782215, 1782457, 1785350, 1788435, 1791179, 1791277, 1791309, 1792037, 1792412, 1794523, 1800624, 1803500, 1804280, 1804765, 1805683, 1806096, 1812446, 1815916, 1817630, 1819339, 1819450, 1820063, 1821069, 1823158, 1824860, 1825093, 1825523, 1828196, 1829131, 1829513, 1830174, 1832172, 1834053, 1834871, 1835132, 1835689, 1837597, 1838010, 1838269, 1838486, 1839073, 1839746, 1842142, 1845693, 1846776, 1851270, 1852623, 1853602, 1855060, 1862422, 1862652, 1864071, 1865113, 1865291, 1865799, 1866579, 1866785, 1867152, 1869162, 1869318, 1870184, 1873384, 1877206, 1878774, 1879399, 1881868, 1882402, 1886970, 1887198, 1888342, 1892312, 1894098, 1896830, 1897984, 1898064, 1900947, 1903663, 1908533, 1916328, 1918491, 1919907, 1922517, 1922874, 1926155, 1928220, 1929576, 1930053, 1940754, 1946527, 1951934, 1951951, 1953416, 1954512, 1960205, 1961268, 1965428, 1966235, 1967044, 1969410, 1970026, 1979513, 1980711, 1982649, 1982722, 1983077, 1983307, 1983615, 1984245, 1986069, 1986177, 1988710, 1989952, 1991320, 1995745, 1996350, 1996750, 2015501, 2015746, 2017170, 2018397, 2024125, 2025088, 2025552, 2027653, 2031783, 2032749, 2035245, 2036135, 2037459, 2039222, 2044508, 2048423, 2049556, 2051142, 2053042, 2059854, 2060010, 2061805, 2062223, 2063492, 2064756, 2064874, 2067210, 2067768, 2071786, 2074932, 2077486, 2080044, 2080948, 2082726, 2083227, 2085469, 2088451, 2091973, 2097334, 2101393, 2107517, 2108076, 2109792, 2109844, 2113806, 2114144, 2115475, 2115515, 2115661, 2115829, 2115885, 2117010, 2117792, 2117915, 2118301, 2118401, 2118768, 2118821, 2121263, 2121662, 2128299, 2128610, 2130822, 2130868, 2130908, 2131491, 2132666, 2132724, 2137863, 2139355, 2141713, 2141867, 2143548, 2143609, 2143770, 2145324, 2146460, 2154785, 2169867, 2171273, 2173193, 2173859, 2177863, 2179252, 2180684, 2182157, 2182953, 2183448, 2183730, 2187527, 2188068, 2190858, 2196540, 2196668, 2209271, 2215644, 2217800, 2226043, 2229447, 2229692, 2229823, 2230789, 2231307, 2236967, 2239045, 2245377, 2248380, 2251748, 2251981, 2253237, 2254195, 2258689, 2260208, 2262246, 2263489, 2265146, 2265283, 2267078, 2267659, 2281766, 2284240, 2288380, 2292452, 2293422, 2294237, 2295217, 2296295, 2299626, 2301362, 2302227, 2312117, 2312644, 2313963, 2314619, 2314911, 2315978, 2327108, 2335397, 2336892, 2337250, 2337641, 2340708, 2344763, 2345086, 2346130, 2347760, 2348698, 2354456, 2355664, 2367866, 2368879, 2369090, 2370401, 2370763, 2378672, 2387521, 2387631, 2388325, 2388884, 2389448, 2389953, 2392977, 2394146, 2394176, 2395168, 2398820, 2401635, 2402783, 2403003, 2403964, 2404466, 2405222, 2407845, 2410472, 2410487, 2415497, 2418874, 2421685, 2421731, 2426061, 2426245, 2426472, 2437022, 2440555, 2441127, 2442024, 2444971, 2446847, 2451840, 2453026, 2454635, 2455322, 2460804, 2463390, 2479411, 2480120, 2483161, 2486108, 2486182, 2488952, 2493621, 2494719, 2497340, 2499607, 2504528, 2509218, 2510850, 2512417, 2512616, 2512671, 2512799, 2513495, 2514092, 2514158, 2514186, 2514226, 2514412, 2514492, 2515376, 2515716, 2515780, 2517866, 2518485, 2522297, 2530627, 2544602, 2546637, 2547501, 2553963, 2553978, 2554254, 2559605, 2560163, 2563714, 2566614, 2568868, 2570969, 2575293, 2578415, 2578609, 2580478, 2581124, 2590026, 2595325, 2602629, 2606689, 2607401, 2609261, 2610717, 2611270, 2612177, 2613968, 2613991, 2614721, 2617334, 2619667, 2628935, 2630099, 2631482, 2631959, 2633677, 2637226, 2637579, 2637689, 2637920, 2640595, 2644036, 2645819, 2652607, 2652627, 2653065, 2657008, 2657435, 2658193, 2659713, 2662389, 2668544, 2669121, 2670228, 2670782, 2671082, 2672817, 2673854, 2674431, 2674913, 2675400, 2676078, 2676974, 2676986, 2678522, 2685297, 2686509, 2689756, 2691752, 2691847, 2696359, 2697251, 2700081, 2700940, 2710254, 2713872, 2713897, 2715739, 2716478, 2718387, 2724141, 2724908, 2726040, 2739568, 2742218, 2747243, 2747359, 2756947, 2757089, 2759033, 2760534, 2760998, 2764251, 2765830, 2769016, 2775461, 2776967, 2777944, 2780351, 2780775, 2781022, 2782230, 2783309, 2785971, 2787151, 2788362, 2789254, 2790387, 2790410, 2791284, 2794275, 2795050, 2798232, 2798379, 2799291, 2801071, 2803911, 2804406, 2808818, 2811791, 2816866, 2818081, 2820596, 2824087, 2824386, 2824516, 2824913, 2825283, 2825290, 2829998, 2832999, 2835565, 2836222, 2838112, 2840190, 2841492, 2846231, 2846850, 2850549, 2850575, 2858490, 2859017, 2865204, 2865324, 2868064, 2870233, 2871123, 2873035, 2874213, 2874823, 2875900, 2878772, 2888533, 2896136, 2903523, 2905881, 2907188, 2907298, 2909924, 2912319, 2923335, 2925134, 2927797, 2928939, 2929192, 2930403, 2936645, 2940631, 2947836, 2948495, 2950279, 2952717, 2952806, 2956360, 2959362, 2962326, 2963327, 2970714, 2973592, 2975944, 2976931, 2981530, 2982680, 2983951, 2985514, 2990875, 2991786, 2999195, 2999207, 3001494, 3001919, 3007415, 3007767, 3020285, 3032087, 3033642, 3041605, 3043690, 3047035, 3047650, 3056908, 3057677, 3063690, 3067154, 3068010, 3075507, 3075642, 3079149, 3081335, 3089826, 3092883, 3097696, 3101778, 3105888, 3108964, 3113595, 3113982, 3115642, 3119828, 3119956, 3123449, 3125143, 3127806, 3137317, 3137602, 3139734, 3141820, 3142810, 3145000, 3153074, 3153898, 3154500, 3158143, 3159911, 3161583, 3166557, 3168337, 3170006, 3175332, 3176106, 3176447, 3177629, 3186379, 3189501, 3191576, 3193571, 3195985, 3201228, 3201331, 3204628, 3205721, 3208357, 3211029, 3220319, 3222322, 3222999, 3228531, 3229927, 3239668, 3240092, 3243610, 3262791, 3276027, 3279457, 3283965, 3284620, 3293643, 3294817, 3294931, 3297399, 3300769, 3302342, 3306270, 3310069, 3315310, 3316099, 3321251, 3324091, 3333180, 3334747, 3335713, 3335950, 3337667, 3344255, 3344391, 3356540, 3359885, 3361967, 3363267, 3363556, 3374574, 3378409, 3389381, 3412869, 3418212, 3418331, 3419034, 3424027, 3426731, 3430600, 3431103, 3431662, 3438352, 3439198, 3443815, 3457670, 3458145, 3461699, 3464785, 3466439, 3471973, 3472036, 3472797, 3474092, 3474653, 3479664, 3479874, 3482060, 3482637, 3485383, 3486964, 3492065, 3492657, 3494737, 3495839, 3496350, 3496778, 3499407, 3501612, 3509547, 3510474, 3512289, 3513036, 3514222, 3516403, 3517053, 3517409, 3522326, 3522511, 3523355, 3523419, 3524836, 3531155, 3533664, 3534225, 3534409, 3537600, 3541124, 3542327, 3543480, 3548997, 3550011, 3551338, 3551691, 3552893, 3553000, 3555377, 3555676, 3556967, 3559230, 3560408, 3566442, 3566841, 3567291, 3567414, 3567514, 3575629, 3578811, 3578832, 3579257, 3585210, 3593395, 3598979, 3602619, 3604982, 3614179, 3616138, 3618875, 3621498, 3622203, 3627982, 3630907, 3636161, 3640058, 3641589, 3647824, 3648924, 3649908, 3658329, 3658346, 3658458, 3663169, 3665686, 3667286, 3673262, 3680952, 3687077, 3687392, 3688036, 3690350, 3700434, 3703708, 3703970, 3704921, 3705154, 3705362, 3719998, 3722766, 3724040, 3724936, 3725278, 3729425, 3730470, 3732545, 3734190, 3736747, 3738033, 3742803, 3744005, 3744808, 3749821, 3756964, 3758200, 3758557, 3759236, 3759377, 3762316, 3762986, 3767300, 3775532, 3776825, 3777776, 3780381, 3781288, 3783409, 3787335, 3787839, 3789290, 3789713, 3790516, 3795516, 3795547, 3796332, 3797443, 3803154, 3803789, 3803808, 3804873, 3804929, 3805497, 3805977, 3810030, 3816363, 3818425, 3820671, 3830788, 3831718, 3832102, 3833071, 3833762, 3834246, 3835026, 3835056, 3835611, 3835801, 3837155, 3839573, 3841216, 3842441, 3845115, 3845339, 3847660, 3858441, 3858830, 3859961, 3866642, 3878333, 3879168, 3880140, 3898314, 3900386, 3904166, 3905927, 3910295, 3914576, 3914709, 3916577, 3918664, 3920385, 3928266, 3929060, 3935764, 3941874, 3942389, 3942982, 3949330, 3953802, 3957159, 3964007, 3967404, 3970169, 3970969, 3976253, 3977539, 3980898, 3985099, 3987134, 3990787, 3990865, 3990915, 3991188, 3992020, 3992162, 4002953, 4003852, 4008596, 4010854, 4014604, 4017465, 4022654, 4023500, 4027773, 4029771, 4035471, 4039020, 4039888, 4046258, 4051960, 4060630, 4062082, 4063603, 4066997, 4070137, 4070702, 4079142, 4083316, 4086589, 4100320, 4101116, 4105251, 4107386, 4108321, 4111713, 4111916, 4112499, 4115065, 4124215, 4124780, 4127435, 4132830, 4136167, 4136365, 4137676, 4147552, 4155348, 4157690, 4159330, 4161293, 4162915, 4163814, 4165436, 4165995, 4172003, 4174765, 4176210, 4182510, 4187911, 4187994, 4189245, 4191011, 4202311, 4208451, 4211279, 4212062, 4212210, 4213940, 4214808, 4218661, 4220193, 4223135, 4224792, 4227662, 4227692, 4227877, 4233000, 4235440, 4235877, 4236352, 4239077, 4240399, 4240601, 4241294, 4251068, 4252031, 4252539, 4255332, 4255946, 4256504, 4258218, 4258756, 4259441, 4263629, 4267114, 4267728, 4268490, 4268949, 4271790, 4272209, 4273613, 4273908, 4278913, 4283381, 4300750, 4301766, 4301894, 4304688, 4305050, 4307903, 4315528, 4316134, 4318071, 4320785, 4321851, 4324456, 4324910, 4326263, 4326454, 4332470, 4338065, 4339987, 4342173, 4343190, 4346578, 4348335, 4359190, 4359774, 4361017, 4361395, 4364542, 4364693, 4365898, 4367159, 4368709, 4381619, 4385825, 4387346, 4388436, 4391886, 4398918, 4405909, 4407380, 4411724, 4412962, 4413987, 4423638, 4425784, 4427998, 4432017, 4432630, 4437506, 4439493, 4439804, 4444055, 4444971, 4445665, 4449772, 4455380, 4456261, 4461152, 4461204, 4475391, 4477271, 4482362, 4485120, 4485847, 4486087, 4487329, 4488267, 4491415, 4498578, 4501072, 4503235, 4503325, 4504325, 4504778, 4514270, 4520504, 4521895, 4523711, 4526190, 4527767, 4530023, 4531943, 4534916, 4541490, 4543142, 4544282, 4549204, 4550348, 4556197, 4558728, 4564194, 4564933, 4571319, 4572043, 4573657, 4578985, 4579060, 4579343, 4579575, 4580838, 4585066, 4585711, 4587239, 4588003, 4589603, 4593718, 4594622, 4595030, 4597571, 4599005, 4599145, 4600997, 4602219, 4607711, 4607869, 4608302, 4609592, 4610412, 4620291, 4632593, 4636089, 4639641, 4642329, 4645875, 4647003, 4648529, 4653627, 4657171, 4659410, 4667915, 4675453, 4682533, 4683871, 4685809, 4686195, 4690079, 4692310, 4693383, 4693834, 4694966, 4695190, 4697358, 4699296, 4700436, 4701220, 4701658, 4703930, 4706721, 4707465, 4708962, 4714797, 4717280, 4717806, 4718618, 4723797, 4723927, 4724247, 4724793, 4725377, 4725440, 4726623, 4728720, 4733608, 4734610, 4738703, 4741334, 4745022, 4750760, 4754574, 4754903, 4764719, 4773672, 4775118, 4776390, 4778231, 4779119, 4780579, 4781184, 4782932, 4785430, 4788017, 4789988, 4790949, 4792984, 4793332, 4793506, 4795594, 4796476, 4800972, 4803105, 4803731, 4805410, 4806066, 4808150, 4808882, 4810358, 4811008, 4812384, 4812422, 4812626, 4813899, 4814986, 4816031, 4817579, 4821096, 4821963, 4837595, 4857484, 4858250, 4860191, 4863248, 4863608, 4863645, 4863920, 4864900, 4865685, 4866582, 4867029, 4867936, 4872118, 4874828, 4882145, 4882865, 4886424, 4888078, 4888406, 4890437, 4893144, 4894302, 4896650, 4897013, 4901809, 4903969, 4907036, 4909756, 4909878, 4911082, 4911519, 4912516, 4916704, 4917155, 4917464, 4920698, 4921336, 4921440, 4921552, 4924378, 4932146, 4936637, 4938386, 4939096, 4940917, 4944708, 4946392, 4963002, 4970372, 4970800, 4972981, 4973642, 4978317, 4980745, 4986792, 4990500, 4994458, 4994531, 4995723, 4996838, 5005168, 5013004, 5020362, 5021938, 5022200, 5028144, 5032434, 5034333, 5038360, 5041127, 5042290, 5042963, 5043396, 5044150, 5046922, 5049659, 5061924, 5070173, 5070706, 5070927, 5071236, 5071663, 5084881, 5086781, 5097506, 5100350, 5101193, 5104588, 5105138, 5105551, 5106110, 5108327, 5112671, 5121779, 5121862, 5128030, 5133765, 5135681, 5141940, 5143155, 5147109, 5149047, 5151648, 5159352, 5160540, 5160926, 5163709, 5164325, 5167895, 5171821, 5175043, 5175623, 5178004, 5178879, 5179708, 5183520, 5193178, 5194256, 5195578, 5196907, 5198289, 5199657, 5201159, 5204669, 5205130, 5205182, 5206109, 5207909, 5210215, 5212669, 5215521, 5218179, 5221552, 5223699, 5223765, 5225846, 5226144, 5229055, 5233772, 5233791, 5240271, 5243531, 5248725, 5250607, 5250620, 5250975, 5266116, 5275259, 5277765, 5278809, 5281639, 5282505, 5292112, 5293064, 5294697, 5296024, 5296865, 5299312, 5302724, 5303756, 5306717, 5308420, 5312173, 5312401, 5312633, 5313336, 5313473, 5313713, 5317415, 5318433, 5320445, 5321798, 5326898, 5327108, 5327208, 5332448, 5343080, 5344620, 5345825, 5348916, 5349782, 5351172, 5355263, 5372244, 5373487, 5374667, 5376596, 5376766, 5377519, 5385439, 5386219, 5389318, 5392477, 5392759, 5395841, 5396402, 5400125, 5403022, 5406999, 5410743, 5410988, 5411459, 5411906, 5417704, 5419513, 5425402, 5426534, 5427456, 5428519, 5429528, 5434093, 5436914, 5438659, 5440247, 5444281, 5451167, 5457785, 5458401, 5461975, 5469148, 5469767, 5470090, 5471987, 5472147, 5473118, 5474101, 5474804, 5478413, 5478522, 5479140, 5488730, 5490769, 5490778, 5494873, 5494987, 5499555, 5508645, 5508867, 5515613, 5516276, 5519759, 5522420, 5524866, 5526456, 5531889, 5536600, 5545891, 5546056, 5546682, 5549292, 5549778, 5549911, 5554333, 5555972, 5557975, 5560128, 5566333, 5569750, 5570612, 5571695, 5589113, 5591502, 5592984, 5594467, 5595566, 5596445, 5598776, 5599403, 5601601, 5601680, 5604041, 5604384, 5605210, 5610752, 5613794, 5617435, 5617436, 5617934, 5618421, 5622189, 5630001, 5631328, 5638913, 5644752, 5653968, 5659426, 5659487, 5659562, 5661477, 5661598, 5669669, 5670707, 5673449, 5673729, 5678031, 5679214, 5680468, 5684034, 5691406, 5691546, 5696556, 5698470, 5699878, 5709219, 5716084, 5716801, 5718960, 5727776, 5731752, 5740712, 5746142, 5746942, 5747107, 5748429, 5754834, 5756589, 5771220, 5778851, 5779301, 5780298, 5780761, 5781728, 5785051, 5793012, 5798136, 5799772, 5800871, 5803392, 5806905, 5807457, 5809689, 5810533, 5812038, 5821490, 5824971, 5825306, 5826709, 5826900, 5826958, 5829863, 5831301, 5833653, 5833825, 5833888, 5841593, 5844572, 5846754, 5848734, 5851917, 5852286, 5853346, 5854655, 5857632, 5860935, 5863650, 5872708, 5875832, 5894326, 5895179, 5908748, 5911218, 5917254, 5919839, 5921918, 5923200, 5924857, 5925043, 5941013, 5945245, 5947607, 5950764, 5957978, 5959991, 5961595, 5961786, 5962099, 5962652, 5967776, 5969492, 5970238, 5973926, 5975827, 5977470, 5978823, 5986101, 5987741, 5988895, 5991366, 6002495, 6005561, 6007749, 6016604, 6021643, 6024007, 6024614, 6024784, 6024940, 6026635, 6029485, 6033632, 6041782, 6044359, 6046997, 6050005, 6050681, 6052290, 6055498, 6057066, 6058704, 6061003, 6074204, 6077958, 6086787, 6091409, 6098935, 6101120, 6101255, 6104222, 6114396, 6114863, 6116403, 6119321, 6122343, 6123259, 6126460, 6126648, 6132014, 6144218, 6146719, 6148174, 6159581, 6176615, 6178182, 6178845, 6179226, 6180590, 6186382, 6187999, 6190165, 6198425, 6199330, 6202339, 6203853, 6205827, 6206370, 6207740, 6209782, 6210682, 6216479, 6217359, 6219557, 6223749, 6226264, 6226733, 6234418, 6236744, 6237129, 6237906, 6237966, 6240983, 6241129, 6243517, 6243951, 6244477, 6246077, 6248524, 6253459, 6253627, 6257596, 6262149, 6263708, 6264757, 6265534, 6271391, 6275284, 6282620, 6286166, 6288559, 6296284, 6310412, 6316166, 6318133, 6322307, 6324238, 6325014, 6327172, 6329080, 6329514, 6332215, 6332678, 6333251, 6340903, 6348485, 6353364, 6353958, 6360996, 6363863, 6368653, 6381527, 6385247, 6385334, 6388195, 6392668, 6394513, 6395232, 6395575, 6396582, 6397749, 6407477, 6407540, 6408147, 6408433, 6411509, 6412008, 6424006, 6427788, 6428756, 6430963, 6432178, 6434840, 6437115, 6450551, 6452858, 6455448, 6455454, 6461754, 6462869, 6463723, 6465687, 6466181, 6469369, 6472076, 6473581, 6474229, 6475436, 6485628, 6491509, 6492967, 6497417, 6504842, 6505651, 6507881, 6508152, 6521874, 6522721, 6528968, 6532996, 6535157, 6538146, 6541481, 6543716, 6545016, 6549218, 6549874, 6550161, 6551514, 6553645, 6560734, 6562237, 6562294, 6577298, 6584342, 6584848, 6585349, 6586468, 6587044, 6587241, 6600062, 6600359, 6603527, 6606784, 6612813, 6614616, 6619645, 6625290, 6630496, 6635219, 6641451, 6648351, 6649792, 6656467, 6660207, 6677158, 6687752, 6690894, 6691324, 6696204, 6698316, 6703227, 6703614, 6719025, 6720318, 6721658, 6727436, 6732031, 6739693, 6740804, 6741503, 6745615, 6745993, 6753396, 6753866, 6761813, 6765868, 6767703, 6770567, 6772264, 6774193, 6778625, 6779845, 6781537, 6788013, 6788959, 6790656, 6791129, 6793650, 6795757, 6796228, 6797147, 6798552, 6801388, 6804278, 6804999, 6805097, 6809437, 6810468, 6813779, 6814570, 6818258, 6818614, 6826846, 6832274, 6853605, 6859667, 6860191, 6862150, 6865325, 6865482, 6866932, 6867192, 6872908, 6873751, 6875004, 6880488, 6884788, 6888806, 6899641, 6901733, 6902992, 6904157, 6907191, 6913085, 6913105, 6913897, 6923889, 6936355, 6947883, 6951493, 6954494, 6958689, 6959078, 6959629, 6965828, 6972991, 6979052, 6980813, 6991527, 6994608, 7013176, 7015624, 7019325, 7025066, 7034773, 7035862, 7037523, 7038917, 7044145, 7047913, 7050418, 7054816, 7055616, 7057514, 7057540, 7062960, 7066755, 7067122, 7069475, 7074951, 7076627, 7079265, 7081909, 7084014, 7084127, 7085147, 7087750, 7088542, 7088945, 7092024, 7101438, 7104968, 7105758, 7106100, 7111851, 7114388, 7114723, 7115398, 7139108, 7139334, 7144283, 7153205, 7162719, 7162869, 7165714, 7167104, 7168329, 7171562, 7175719, 7178510, 7181265, 7181534, 7187529, 7190095, 7193066, 7195367, 7196221, 7198452, 7199337, 7204225, 7207010, 7207688, 7213503, 7217305, 7219743, 7223699, 7235026, 7235291, 7236344, 7236602, 7236832, 7237486, 7253727, 7253856, 7257578, 7260125, 7260144, 7264701, 7267113, 7269055, 7270272, 7272359, 7272696, 7276733, 7282163, 7282685, 7287456, 7288683, 7289279, 7308442, 7309814, 7311006, 7313441, 7315725, 7317741, 7318774, 7331786, 7332618, 7332704, 7333976, 7335300, 7339107, 7339267, 7340853, 7343299, 7349424, 7355155, 7361310, 7378348, 7379457, 7389429, 7390814, 7394420, 7395910, 7403363, 7404711, 7411413, 7411986, 7414920, 7418711, 7421159, 7424323, 7424788, 7425466, 7429868, 7431218, 7444582, 7447994, 7451398, 7456924, 7459013, 7462955, 7463375, 7463621, 7472343, 7476569, 7477106, 7478267, 7481215, 7497739, 7498302, 7498966, 7505656, 7505796, 7512216, 7515133, 7517683, 7517776, 7528655, 7529960, 7535456, 7545239, 7548307, 7556974, 7561004, 7563563, 7563659, 7563763, 7569174, 7574499, 7575110, 7575369, 7579093, 7598795, 7599928, 7600109, 7600817, 7604494, 7606420, 7608444, 7609074, 7610555, 7611346, 7611469, 7612328, 7613563, 7615466, 7633734, 7635171, 7636650, 7637533, 7638520, 7638892, 7657199, 7660121, 7666587, 7670914, 7674946, 7694297, 7701111, 7701927, 7702403, 7702844, 7703164, 7705233, 7710580, 7713140, 7723000, 7725159, 7728951, 7729448, 7732106, 7737396, 7738804, 7741240, 7744142, 7752021, 7765332, 7767287, 7772353, 7779301, 7780980, 7782629, 7783488, 7783965, 7792953, 7795911, 7796899, 7798419, 7802954, 7811781, 7812140, 7817050, 7817217, 7820328, 7821179, 7821776, 7825254, 7828491, 7829652, 7830521, 7831028, 7848271, 7857150, 7858504, 7861364, 7862495, 7866773, 7867174, 7868666, 7869102, 7873575, 7875229, 7875912, 7876206, 7877402, 7890027, 7894479, 7904815, 7911769, 7911794, 7915512, 7920763, 7921334, 7926998, 7930077, 7938374, 7938885, 7938906, 7939242, 7940016, 7946086, 7948343, 7950431, 7951137, 7953847, 7954803, 7956936, 7967147, 7973162, 7976490, 7976927, 7978198, 7984895, 7996728, 7999306, 8000046, 8000341, 8002134, 8002364, 8012174, 8015920, 8022974, 8025018, 8026715, 8028906, 8037545, 8049760, 8050741, 8059124, 8059163, 8062578, 8068361, 8069918, 8073792, 8081239, 8083856, 8083991, 8088487, 8099434, 8104398, 8110575, 8118671, 8120498, 8128820, 8130721, 8131705, 8132486, 8134142, 8138081, 8139470, 8139534, 8139899, 8142628, 8145391, 8148005, 8150850, 8151566, 8156080, 8156715, 8162240, 8164928, 8175063, 8182233, 8190293, 8203281, 8203597, 8223981, 8224971, 8226558, 8227916, 8228768, 8234572, 8235410, 8258393, 8263131, 8274346, 8276165, 8281635, 8305302, 8347914, 8353493, 8376796, 8377409, 8378566, 8381254, 8382370, 8400251, 8400458, 8413551, 8431573, 8432380, 8447124, 8485173, 8486219, 8494970, 8495659, 8498146, 8525546, 8534251, 8536222, 8549120, 8555135, 8564180, 8568677, 8574037, 8577660, 8579956, 8580374, 8582643, 8584170, 8585004, 8597522, 8597691, 8598394, 8600820, 8644086, 8646657, 8650996, 8656794, 8662139, 8663789, 8675384, 8675589, 8678580, 8678849, 8689302, 8692574, 8705083, 8705144, 8707354, 8711631, 8713595, 8722594, 8725279, 8727619, 8733735, 8742157, 8744503, 8745162, 8745352, 8745628, 8745888, 8747247, 8749288, 8750122, 8751962, 8766412, 8769784, 8770288, 8770731, 8772754, 8781245, 8802916, 8804356, 8815127, 8824256, 8827960, 8850404, 8853618, 8856883, 8861859, 8867428, 8869608, 8876590, 8891460, 8891684, 8903787, 8911179, 8913492, 8916255, 8918121, 8921900, 8922102, 8933678, 8935727, 8936443, 8936569, 8957722, 8960628, 8965012, 8977004, 8979850, 8986807, 8986951, 8988808, 9013172, 9022565, 9024633, 9029366, 9051516, 9056272, 9064811, 9065100, 9068963, 9076916, 9079242, 9084457, 9090597, 9095099, 9099550, 9100329, 9117350, 9120241, 9137566, 9138407, 9138531, 9140876, 9142202, 9143296, 9143507, 9147789, 9153784, 9155990, 9174350, 9174665, 9174968, 9181030, 9184118, 9188113, 9195444, 9198097, 9198271, 9200366, 9200696, 9206426, 9209255, 9225017, 9231308, 9231906, 9248197, 9261793, 9285607, 9288483, 9294590, 9300212, 9303388, 9305300, 9330463, 9330988, 9332547, 9341990, 9342005, 9344425, 9353702, 9353953, 9366429, 9377459, 9377577, 9384716, 9388566, 9392213, 9393349, 9393666, 9394114, 9394133, 9394138, 9400483, 9405319, 9407149, 9411611, 9413701, 9415459, 9417518, 9425061, 9440523, 9442290, 9442366, 9444688, 9445556, 9445577, 9445578, 9450603, 9454181, 9460104, 9464003, 9470262, 9476328, 9484775, 9517845, 9531463, 9531528, 9538107, 9541648, 9559120, 9562214, 9591321, 9591540, 9593498, 9595767, 9596951, 9603002, 9616886, 9618154, 9619356, 9622454, 9625061, 9637706, 9637937, 9639674, 9642660, 9642721, 9648676, 9657006, 9668315, 9670967, 9673702, 9678195, 9678591, 9679378, 9679637, 9696651, 9723694, 9738587, 9744092, 9753172, 9758033, 9786180, 9806890, 9807483, 9820476, 9822065, 9859152]
Cluster 2 (Cluster 2):
Users: [4754125]

K means Evaluation

inertias = []
for k_value in range(1, 6):
    model = KMeans(n_clusters=k_value, random_state=0)
    model.fit(user_data_array)
    inertias.append(model.inertia_)

plt.plot(range(1, 6), inertias, marker='o')
plt.xlabel('Number of Clusters (K)')
plt.ylabel('Inertia')
plt.title('Elbow Method for Optimal K')

C:\Users\abhiv\AppData\Local\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:870: FutureWarning: The default value of `n_init` will change from 10 to 'auto' in 1.4. Set the value of `n_init` explicitly to suppress the warning
  warnings.warn(
C:\Users\abhiv\AppData\Local\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:870: FutureWarning: The default value of `n_init` will change from 10 to 'auto' in 1.4. Set the value of `n_init` explicitly to suppress the warning
  warnings.warn(
C:\Users\abhiv\AppData\Local\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:870: FutureWarning: The default value of `n_init` will change from 10 to 'auto' in 1.4. Set the value of `n_init` explicitly to suppress the warning
  warnings.warn(
C:\Users\abhiv\AppData\Local\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:870: FutureWarning: The default value of `n_init` will change from 10 to 'auto' in 1.4. Set the value of `n_init` explicitly to suppress the warning
  warnings.warn(
C:\Users\abhiv\AppData\Local\anaconda3\Lib\site-packages\sklearn\cluster\_kmeans.py:870: FutureWarning: The default value of `n_init` will change from 10 to 'auto' in 1.4. Set the value of `n_init` explicitly to suppress the warning
  warnings.warn(





Text(0.5, 1.0, 'Elbow Method for Optimal K')

from sklearn.metrics import silhouette_score

silhouette_avg = silhouette_score(user_data_array, y_pred)
print(f"Silhouette Score: {silhouette_avg}")

Silhouette Score: 0.9098500527957909

K means Labeling Analysis

user_data_3d_turnover = user_data_3d['Turnover']

# Create a new DataFrame to store the aggregated features
agguser_allfeature = pd.DataFrame()

#['Turnover', 'Hold', 'NumberofBets', 'YearofBirth', 'Interventiontype_first']

# Calculate the mean of each row and add it as a new column
agguser_allfeature['Mean_Turnover'] = user_data_3d_turnover.mean(axis=1)
agguser_allfeature['Mean_Hold'] = user_data_3d['Hold'].mean(axis=1)  
agguser_allfeature['Mean_NumberofBets'] = user_data_3d['NumberofBets'].mean(axis=1)  
agguser_allfeature['Mean_YearofBirth'] = user_data_3d['Age_until_2010'].mean(axis=1)  
agguser_allfeature['Mean_Interventiontype_first'] = user_data_3d['Interventiontype_first'].mean(axis=1)
agguser_allfeature['Cluster'] = y_pred


# Set the index of agguser_allfeaturue to match user_data_3d_turnover
agguser_allfeature.index = user_data_3d_turnover.index.tolist()

# Display the updated DataFrame
print(agguser_allfeature)

         Mean_Turnover  Mean_Hold  Mean_NumberofBets  Mean_YearofBirth  \
31965        29.325776   4.097601           3.243651         12.808167   
32639         0.001836   0.001836           0.000343          0.014070   
36822         0.008236  -0.008476           0.002059          0.027454   
36916         0.027111   0.027111           0.001030          0.028140   
74438         0.383665   0.150206           0.008236          0.180165   
...                ...        ...                ...               ...   
9806890       0.028240  -0.000159           0.002745          0.009952   
9807483       0.102132  -0.015731           0.003775          0.009266   
9820476       0.030298  -0.001102           0.001030          0.008579   
9822065       0.107430   0.016380           0.058339          0.354839   
9859152       0.004461   0.004461           0.001716          0.009609   

         Mean_Interventiontype_first  Cluster  
31965                       2.627316        1  
32639                       0.000000        1  
36822                       0.000000        1  
36916                       0.004118        1  
74438                       0.000000        1  
...                              ...      ...  
9806890                     0.000000        1  
9807483                     0.000343        1  
9820476                     0.004461        1  
9822065                     0.098147        1  
9859152                     0.000000        1  

[3161 rows x 6 columns]

from pandas.plotting import parallel_coordinates
import seaborn as sns

# Cast the index to integers
agguser_allfeature.index = agguser_allfeature.index.astype(int)

# Create a scatter plot
plt.figure(figsize=(10, 6))
sns.scatterplot(data=agguser_allfeature, x='Mean_Turnover', y='Mean_Hold', hue='Cluster', palette='viridis', alpha=0.7)
plt.xlabel('Mean_Turnover')
plt.ylabel('Mean_Hold')
plt.title('Cluster Visualization based on Mean Turnover and Mean Hold')
#plt.legend(title='Cluster', loc='upper right')
legend_labels = ['Moderate Problem Gamblers','Early Players','Problem Gamblers']# [f'Cluster {info["cluster_num"]}' for info in cluster_dict.values()]
legend = plt.legend(handles=scatter.legend_elements()[0], title='Cluster', labels=legend_labels)


plt.show()

# Create a scatter plot
plt.figure(figsize=(10, 6))
sns.scatterplot(data=agguser_allfeature, x='Mean_Hold', y='Mean_NumberofBets', hue='Cluster', palette='viridis', alpha=0.7)
plt.xlabel('Mean_Hold')
plt.ylabel('Mean_NumberofBets')
plt.title('Cluster Visualization based on Mean Hold and Mean Number of Bets')
#plt.legend(title='Cluster', loc='upper right')
legend_labels = ['Moderate Problem Gamblers','Early Players','Problem Gamblers']# [f'Cluster {info["cluster_num"]}' for info in cluster_dict.values()]
legend = plt.legend(handles=scatter.legend_elements()[0], title='Cluster', labels=legend_labels)

plt.show()

from pandas.plotting import parallel_coordinates

# Define custom colors for each cluster
custom_colors = [ 'teal','purple', 'yellow']

# Create a parallel coordinates plot with custom colors
plt.figure(figsize=(12, 6))
parallel_coordinates(agguser_allfeature, 'Cluster', colormap='viridis', alpha=0.7, color=custom_colors)
plt.title('Parallel Coordinates Plot for Cluster Visualization')

# Add a custom legend
legend_labels = ['Moderate Problem Gamblers','Early Players','Problem Gamblers']# [f'Cluster {info["cluster_num"]}' for info in cluster_dict.values()]
legend = plt.legend(handles=scatter.legend_elements()[0], title='Cluster', labels=legend_labels)

plt.show()

C:\Users\abhiv\AppData\Local\Temp\ipykernel_14916\1398724544.py:8: UserWarning: 'color' and 'colormap' cannot be used simultaneously. Using 'color'
  parallel_coordinates(agguser_allfeature, 'Cluster', colormap='viridis', alpha=0.7, color=custom_colors)
C:\Users\abhiv\AppData\Local\anaconda3\Lib\site-packages\IPython\core\pylabtools.py:152: UserWarning: Creating legend with loc="best" can be slow with large amounts of data.
  fig.canvas.print_figure(bytes_io, **kw)

### Interactive Slider to understand Clustering using K - means

import numpy as np
import pandas as pd
from sklearn.cluster import KMeans
import matplotlib.pyplot as plt
from ipywidgets import interact

from ipywidgets import interact, DatePicker



# Create a function to plot the clustering results for a specific time frame
def plot_clusters(time_frame):
    print(time_frame)
    start_date = '2002-11-12' #).date()
        # Extract the year from the time_frame and construct a new start_date
    start_year = str(int(time_frame))
    start_date = pd.to_datetime(start_year + '-11-12')
    
    plt.figure(figsize=(8, 6))
    plt.scatter(user_data_3d['NumberofBets'][start_date], user_data_3d['Hold'][start_date], c=y_pred, cmap='viridis')
   # plt.scatter(user_data_array[:, 0], user_data_array[:, 2], c=y_pred, cmap='viridis')
    plt.xlabel('NumberofBets')
    plt.ylabel('Hold')
    plt.title(f'K-means Clustering: NumberofBets vs Hold (Time Frame {time_frame})')
    
   
    # filtered_data_array = user_data_array[time_frame_condition]
    # filtered_y_pred = y_pred[time_frame_condition]
    
    # Scatter plot the filtered data
    # plt.scatter(filtered_data_array[:, 0], filtered_data_array[:, 2], c=filtered_y_pred, cmap='viridis')
    
    plt.show()
    

start_date='2002-11-12'
end_date='2010-11-10'




# Create the interaction using the date pickers
interact(plot_clusters, time_frame=(2002, 2009))

interactive(children=(IntSlider(value=2005, description='time_frame', max=2009, min=2002), Output()), _dom_cla…





<function __main__.plot_clusters(time_frame)>

import matplotlib.pyplot as plt

# List of 10 user IDs to plot
user_ids_to_plot = [31965, 32639, 36822, 36916, 74438, 90746, 91707, 92140, 96950, 99596] 

# Create a subplot for the graph
plt.figure(figsize=(12, 8))
plt.title('Turnover Values for 10 Users')
plt.xlabel('Date')
plt.ylabel('Turnover')
plt.grid(True)

for user_id in user_ids_to_plot:

    
    
    user_data_for_user = user_data_3d.loc[user_id, 'Turnover']
    
    # Extract the turnover values
    turnover_values = user_data_for_user.to_numpy()
    
    # Extract the corresponding date indices
    dates = user_data_for_user.index.get_level_values('Aggregate_Date')
    
    # Plot the turnover values against dates
    plt.plot(dates, turnover_values, marker='o', linestyle='-', label=f'User {user_id}')

# Add a legend to differentiate the users
plt.legend(loc='upper right')

# Rotate the x-axis labels for better visibility
plt.xticks(rotation=45)

# Show the plot
plt.show()

user_data_3d

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead tr th {
    text-align: left;
}

.dataframe thead tr:last-of-type th {
    text-align: right;
}

</style>

	Turnover										...	Event_type_first
Aggregate_Date	2002-11-12	2002-11-13	2002-11-14	2002-11-15	2002-11-16	2002-11-17	2002-11-18	2002-11-19	2002-11-20	2002-11-22	...	2010-11-01	2010-11-02	2010-11-03	2010-11-04	2010-11-05	2010-11-06	2010-11-07	2010-11-08	2010-11-09	2010-11-10
UserID
31965	20.0	0.0	73.18	10.0	163.28	162.34	156.0	3.74	80.0	0.0	...	0.0	0.0	0.0	2.0	0.0	2.0	0.0	0.0	0.0	0.0
32639	0.0	0.0	0.00	0.0	0.00	0.00	0.0	0.00	0.0	0.0	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
36822	0.0	0.0	0.00	0.0	0.00	0.00	0.0	0.00	0.0	0.0	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
36916	0.0	0.0	0.00	0.0	0.00	0.00	0.0	0.00	0.0	0.0	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
74438	0.0	0.0	0.00	0.0	0.00	0.00	0.0	0.00	0.0	0.0	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
9806890	0.0	0.0	0.00	0.0	0.00	0.00	0.0	0.00	0.0	0.0	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
9807483	0.0	0.0	0.00	0.0	0.00	0.00	0.0	0.00	0.0	0.0	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
9820476	0.0	0.0	0.00	0.0	0.00	0.00	0.0	0.00	0.0	0.0	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
9822065	0.0	0.0	0.00	0.0	0.00	0.00	0.0	0.00	0.0	0.0	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
9859152	0.0	0.0	0.00	0.0	0.00	0.00	0.0	0.00	0.0	0.0	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0

3161 rows × 23312 columns

# features to consider : turnover, hold , number of bets for time series prediction

moderate_pg_players= [868583, 1175809, 1411743, 1457496, 1486136, 1662632, 1679490, 1776178, 1790848, 1921204, 2070894, 2150296, 2155065, 2589710, 2704382, 2852203, 3669466, 3852889, 3904422, 3968386, 4006708, 4371320, 4394754, 4412550, 4495603, 4532357, 5106620, 5308271, 5488160, 5660719, 5678852, 5723033, 6158120, 6175402, 6239380, 6283338, 6709379, 6985339, 7192925,4754125]

count = len(moderate_pg_players)
print("Count of moderate addicted players:", count)

Count of moderate addicted players: 40

bv=user_data_3d['Hold'][user_data_3d.index == 868583]
bv.value_counts
#user_data_3d[user_data_3d['UserID']==868583]

<bound method DataFrame.value_counts of Aggregate_Date  2002-11-12  2002-11-13  2002-11-14  2002-11-15  2002-11-16  \
UserID                                                                       
868583                 0.0         0.0         0.0         0.0         0.0   

Aggregate_Date  2002-11-17  2002-11-18  2002-11-19  2002-11-20  2002-11-22  \
UserID                                                                       
868583                 0.0         0.0         0.0         0.0         0.0   

Aggregate_Date  ...  2010-11-01  2010-11-02  2010-11-03  2010-11-04  \
UserID          ...                                                   
868583          ...         0.0         0.0         0.0         0.0   

Aggregate_Date  2010-11-05  2010-11-06  2010-11-07  2010-11-08  2010-11-09  \
UserID                                                                       
868583                 0.0         0.0         0.0         0.0         0.0   

Aggregate_Date  2010-11-10  
UserID                      
868583                 0.0  

[1 rows x 2914 columns]>

Filtering of data based on hypothesis testing for Stationarity

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from statsmodels.tsa.stattools import adfuller

# Create a dictionary to store user IDs, their corresponding 'is_stationary' values, and 'hypothesis_test_result'
is_stationary_data = {'UserID': moderate_pg_players, 'is_stationary': [], 'hypothesis_test_result': []}

# Iterate through the list of user IDs
for user_id in moderate_pg_players:
    # Extract the user's turnover data
    user_turnover = user_data_3d['Turnover'][user_data_3d.index == user_id].values.ravel()
    
    # Perform the ADF test for stationarity
    result = adfuller(user_turnover)
    
    # Check if the time series is stationary based on the p-value
    if result[1] <= 0.05:
        is_stationary = 1  # Stationary
        hypothesis_test_result = 'Reject Null Hypothesis'  # Stationarity is significant
    else:
        is_stationary = 0  # Not stationary
        hypothesis_test_result = 'Fail to Reject Null Hypothesis'  # Stationarity is not significant
    
    # Append the 'is_stationary' and 'hypothesis_test_result' values to the list
    is_stationary_data['is_stationary'].append(is_stationary)
    is_stationary_data['hypothesis_test_result'].append(hypothesis_test_result)

# Create a new DataFrame from the dictionary
is_stationary_df = pd.DataFrame(is_stationary_data)

# Print the new DataFrame
print(is_stationary_df)

     UserID  is_stationary          hypothesis_test_result
0    868583              1          Reject Null Hypothesis
1   1175809              1          Reject Null Hypothesis
2   1411743              1          Reject Null Hypothesis
3   1457496              1          Reject Null Hypothesis
4   1486136              1          Reject Null Hypothesis
5   1662632              1          Reject Null Hypothesis
6   1679490              1          Reject Null Hypothesis
7   1776178              1          Reject Null Hypothesis
8   1790848              1          Reject Null Hypothesis
9   1921204              1          Reject Null Hypothesis
10  2070894              1          Reject Null Hypothesis
11  2150296              1          Reject Null Hypothesis
12  2155065              1          Reject Null Hypothesis
13  2589710              1          Reject Null Hypothesis
14  2704382              1          Reject Null Hypothesis
15  2852203              1          Reject Null Hypothesis
16  3669466              1          Reject Null Hypothesis
17  3852889              1          Reject Null Hypothesis
18  3904422              1          Reject Null Hypothesis
19  3968386              1          Reject Null Hypothesis
20  4006708              1          Reject Null Hypothesis
21  4371320              1          Reject Null Hypothesis
22  4394754              1          Reject Null Hypothesis
23  4412550              1          Reject Null Hypothesis
24  4495603              0  Fail to Reject Null Hypothesis
25  4532357              1          Reject Null Hypothesis
26  5106620              1          Reject Null Hypothesis
27  5308271              1          Reject Null Hypothesis
28  5488160              1          Reject Null Hypothesis
29  5660719              1          Reject Null Hypothesis
30  5678852              1          Reject Null Hypothesis
31  5723033              1          Reject Null Hypothesis
32  6158120              1          Reject Null Hypothesis
33  6175402              1          Reject Null Hypothesis
34  6239380              0  Fail to Reject Null Hypothesis
35  6283338              1          Reject Null Hypothesis
36  6709379              1          Reject Null Hypothesis
37  6985339              1          Reject Null Hypothesis
38  7192925              1          Reject Null Hypothesis
39  4754125              1          Reject Null Hypothesis

# Stationray for Hold feature

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from statsmodels.tsa.stattools import adfuller

# Create a dictionary to store user IDs and their corresponding 'is_stationary' values
is_stationary_data = {'UserID': moderate_pg_players, 'is_stationary_hold': []}

# Iterate through the list of user IDs
for user_id in moderate_pg_players:
    # Extract the user's turnover data
    user_turnover = user_data_3d['Hold'][user_data_3d.index == user_id].values.ravel()
    
    # Perform the ADF user_data_3d_timeseries_stationarity_check
    result = adfuller(user_turnover)
    
    # Check if the time series is stationary based on the p-value
    if result[1] <= 0.05:
        is_stationary = 1  # Stationary
    else:
        is_stationary = 0  # Not stationary
    
    # Append the 'is_stationary' value to the list
    is_stationary_data['is_stationary_hold'].append(is_stationary)

# Create a new DataFrame from the dictionary
is_stationary_df['is_stationary_hold'] = is_stationary_data['is_stationary_hold']

# Print the new DataFrame
#print(is_stationary_df)

#NumberofBets

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from statsmodels.tsa.stattools import adfuller

# Create a dictionary to store user IDs and their corresponding 'is_stationary' values
is_stationary_data = {'UserID': moderate_pg_players, 'is_stationary_NumberofBets': []}

# Iterate through the list of user IDs
for user_id in moderate_pg_players:
    # Extract the user's turnover data
    user_turnover = user_data_3d['NumberofBets'][user_data_3d.index == user_id].values.ravel()
    
    # Perform the ADF user_data_3d_timeseries_stationarity_check
    result = adfuller(user_turnover)
    
    # Check if the time series is stationary based on the p-value
    if result[1] <= 0.05:
        is_stationary = 1  # Stationary
    else:
        is_stationary = 0  # Not stationary
    
    # Append the 'is_stationary' value to the list
    is_stationary_data['is_stationary_NumberofBets'].append(is_stationary)

# Create a new DataFrame from the dictionary
is_stationary_df['is_stationary_NumberofBets'] = is_stationary_data['is_stationary_NumberofBets']

# Print the new DataFrame
#print(is_stationary_df)

#YearofBirth

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from statsmodels.tsa.stattools import adfuller

# Create a dictionary to store user IDs and their corresponding 'is_stationary' values
is_stationary_data = {'UserID': moderate_pg_players, 'is_stationary_Age_until_2010': []}

# Iterate through the list of user IDs
for user_id in moderate_pg_players:
    # Extract the user's turnover data
    user_turnover = user_data_3d['Age_until_2010'][user_data_3d.index == user_id].values.ravel()
    
    # Perform the ADF user_data_3d_timeseries_stationarity_check
    result = adfuller(user_turnover)
    
    # Check if the time series is stationary based on the p-value
    if result[1] <= 0.05:
        is_stationary = 1  # Stationary
    else:
        is_stationary = 0  # Not stationary
    
    # Append the 'is_stationary' value to the list
    is_stationary_data['is_stationary_Age_until_2010'].append(is_stationary)

# Create a new DataFrame from the dictionary
is_stationary_df['is_stationary_Age_until_2010'] = is_stationary_data['is_stationary_Age_until_2010']

# Print the new DataFrame
#print(is_stationary_df)

#Interventiontype_first

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from statsmodels.tsa.stattools import adfuller

# Create a dictionary to store user IDs and their corresponding 'is_stationary' values
is_stationary_data = {'UserID': moderate_pg_players, 'is_stationary_Interventiontype_first': []}

# Iterate through the list of user IDs
for user_id in moderate_pg_players:
    # Extract the user's turnover data
    user_turnover = user_data_3d['Interventiontype_first'][user_data_3d.index == user_id].values.ravel()
    
    # Perform the ADF user_data_3d_timeseries_stationarity_check
    result = adfuller(user_turnover)
    
    # Check if the time series is stationary based on the p-value
    if result[1] <= 0.05:
        is_stationary = 1  # Stationary
    else:
        is_stationary = 0  # Not stationary
    
    # Append the 'is_stationary' value to the list
    is_stationary_data['is_stationary_Interventiontype_first'].append(is_stationary)

# Create a new DataFrame from the dictionary
is_stationary_df['is_stationary_Interventiontype_first'] = is_stationary_data['is_stationary_Interventiontype_first']

# Print the new DataFrame
#print(is_stationary_df)

C:\Users\abhiv\AppData\Local\anaconda3\Lib\site-packages\statsmodels\regression\linear_model.py:940: RuntimeWarning: divide by zero encountered in log
  llf = -nobs2*np.log(2*np.pi) - nobs2*np.log(ssr / nobs) - nobs2
C:\Users\abhiv\AppData\Local\anaconda3\Lib\site-packages\statsmodels\regression\linear_model.py:940: RuntimeWarning: divide by zero encountered in log
  llf = -nobs2*np.log(2*np.pi) - nobs2*np.log(ssr / nobs) - nobs2

is_stationary_df

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

</style>

	UserID	is_stationary	hypothesis_test_result	is_stationary_hold	is_stationary_NumberofBets	is_stationary_Age_until_2010	is_stationary_Interventiontype_first
0	868583	1	Reject Null Hypothesis	1	1	1	1
1	1175809	1	Reject Null Hypothesis	1	1	1	1
2	1411743	1	Reject Null Hypothesis	1	1	1	1
3	1457496	1	Reject Null Hypothesis	1	1	1	1
4	1486136	1	Reject Null Hypothesis	1	1	1	1
5	1662632	1	Reject Null Hypothesis	1	1	1	1
6	1679490	1	Reject Null Hypothesis	1	1	1	1
7	1776178	1	Reject Null Hypothesis	1	1	1	0
8	1790848	1	Reject Null Hypothesis	1	1	1	1
9	1921204	1	Reject Null Hypothesis	1	1	1	1
10	2070894	1	Reject Null Hypothesis	1	1	1	1
11	2150296	1	Reject Null Hypothesis	1	1	1	1
12	2155065	1	Reject Null Hypothesis	1	1	1	1
13	2589710	1	Reject Null Hypothesis	1	1	1	1
14	2704382	1	Reject Null Hypothesis	1	1	1	1
15	2852203	1	Reject Null Hypothesis	1	1	1	1
16	3669466	1	Reject Null Hypothesis	1	1	1	1
17	3852889	1	Reject Null Hypothesis	1	0	0	0
18	3904422	1	Reject Null Hypothesis	1	1	1	0
19	3968386	1	Reject Null Hypothesis	1	1	1	1
20	4006708	1	Reject Null Hypothesis	1	1	1	1
21	4371320	1	Reject Null Hypothesis	1	1	0	0
22	4394754	1	Reject Null Hypothesis	1	1	1	1
23	4412550	1	Reject Null Hypothesis	1	1	0	0
24	4495603	0	Fail to Reject Null Hypothesis	1	1	1	1
25	4532357	1	Reject Null Hypothesis	1	1	1	1
26	5106620	1	Reject Null Hypothesis	1	1	1	1
27	5308271	1	Reject Null Hypothesis	1	1	0	0
28	5488160	1	Reject Null Hypothesis	1	1	1	1
29	5660719	1	Reject Null Hypothesis	1	1	1	1
30	5678852	1	Reject Null Hypothesis	1	1	1	1
31	5723033	1	Reject Null Hypothesis	1	1	1	1
32	6158120	1	Reject Null Hypothesis	1	1	0	0
33	6175402	1	Reject Null Hypothesis	1	1	1	1
34	6239380	0	Fail to Reject Null Hypothesis	1	1	1	1
35	6283338	1	Reject Null Hypothesis	1	1	1	1
36	6709379	1	Reject Null Hypothesis	1	1	1	1
37	6985339	1	Reject Null Hypothesis	1	1	0	0
38	7192925	1	Reject Null Hypothesis	1	1	1	1
39	4754125	1	Reject Null Hypothesis	1	1	1	1

arimausers = is_stationary_df[(is_stationary_df['is_stationary'] == 1) & (is_stationary_df['is_stationary_hold'] == 1) & (is_stationary_df['is_stationary_NumberofBets'] == 1) &  (is_stationary_df['is_stationary_Age_until_2010'] == 1) & (is_stationary_df['is_stationary_Interventiontype_first'] == 1) ]
user_data_3d['Turnover']
x_single_user_turnover=user_data_3d['Turnover'][user_data_3d.index == arimausers['UserID'][1]] #[arimausers['UserID'][1]]

arimausers.reset_index()#[x_single_user_turnover['UserID']=='800']

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

</style>

	index	UserID	is_stationary	hypothesis_test_result	is_stationary_hold	is_stationary_NumberofBets	is_stationary_Age_until_2010	is_stationary_Interventiontype_first
0	0	868583	1	Reject Null Hypothesis	1	1	1	1
1	1	1175809	1	Reject Null Hypothesis	1	1	1	1
2	2	1411743	1	Reject Null Hypothesis	1	1	1	1
3	3	1457496	1	Reject Null Hypothesis	1	1	1	1
4	4	1486136	1	Reject Null Hypothesis	1	1	1	1
5	5	1662632	1	Reject Null Hypothesis	1	1	1	1
6	6	1679490	1	Reject Null Hypothesis	1	1	1	1
7	8	1790848	1	Reject Null Hypothesis	1	1	1	1
8	9	1921204	1	Reject Null Hypothesis	1	1	1	1
9	10	2070894	1	Reject Null Hypothesis	1	1	1	1
10	11	2150296	1	Reject Null Hypothesis	1	1	1	1
11	12	2155065	1	Reject Null Hypothesis	1	1	1	1
12	13	2589710	1	Reject Null Hypothesis	1	1	1	1
13	14	2704382	1	Reject Null Hypothesis	1	1	1	1
14	15	2852203	1	Reject Null Hypothesis	1	1	1	1
15	16	3669466	1	Reject Null Hypothesis	1	1	1	1
16	19	3968386	1	Reject Null Hypothesis	1	1	1	1
17	20	4006708	1	Reject Null Hypothesis	1	1	1	1
18	22	4394754	1	Reject Null Hypothesis	1	1	1	1
19	25	4532357	1	Reject Null Hypothesis	1	1	1	1
20	26	5106620	1	Reject Null Hypothesis	1	1	1	1
21	28	5488160	1	Reject Null Hypothesis	1	1	1	1
22	29	5660719	1	Reject Null Hypothesis	1	1	1	1
23	30	5678852	1	Reject Null Hypothesis	1	1	1	1
24	31	5723033	1	Reject Null Hypothesis	1	1	1	1
25	33	6175402	1	Reject Null Hypothesis	1	1	1	1
26	35	6283338	1	Reject Null Hypothesis	1	1	1	1
27	36	6709379	1	Reject Null Hypothesis	1	1	1	1
28	38	7192925	1	Reject Null Hypothesis	1	1	1	1
29	39	4754125	1	Reject Null Hypothesis	1	1	1	1

print(*arimausers['UserID'])

868583 1175809 1411743 1457496 1486136 1662632 1679490 1790848 1921204 2070894 2150296 2155065 2589710 2704382 2852203 3669466 3968386 4006708 4394754 4532357 5106620 5488160 5660719 5678852 5723033 6175402 6283338 6709379 7192925 4754125

Model Fitting

ARIMA/SARIMA model fitting

from statsmodels.tsa.statespace.sarimax import SARIMAX
from sklearn.preprocessing import MinMaxScaler
import matplotlib.pyplot as plt
import pandas as pd
from sklearn.metrics import mean_squared_error
from statsmodels.tsa.arima.model import ARIMA



# Loop through each user in arimausers
for user_id in arimausers['UserID'].head(10).tolist() + [4754125]:
    x_single_user_turnover = user_data_3d['Turnover'][user_data_3d.index == user_id]
    x_single_user_turnover_ravel = x_single_user_turnover.values.ravel()

    

    frequency = 'D'  # Daily frequency

    start_date='2002-11-12'
    end_date='2010-11-10'



    # Create a time index for the data
    time_index = pd.date_range(start=start_date, periods=len(x_single_user_turnover_ravel), freq=frequency)


    # Scale the data
    scaler = MinMaxScaler(feature_range=(0, 1))
    x_single_user_turnover_ravel_scaled = scaler.fit_transform(x_single_user_turnover_ravel.reshape(-1, 1))

    # Split the data into training and test sets
    split_ratio = 0.97  # 97% for training, 3% for testing
    split_index = int(len(x_single_user_turnover_ravel) * split_ratio)

    train_data = x_single_user_turnover_ravel_scaled[:split_index]
    test_data = x_single_user_turnover_ravel_scaled[split_index:]

    # Create a time index for the data
    time_index_train = pd.date_range(start=start_date, periods=len(train_data), freq=frequency)

    # Create a time index for the data
    time_index_test = pd.date_range(start=time_index_train[-1], periods=len(test_data), freq=frequency)
    
    #ARIMA 
    
    # Define the ARIMA model
    model_arima = ARIMA(train_data, order=(1, 1, 1))

    # Fit the ARIMA model to the training data
    FITmodel_arima = model_arima.fit()

    # Forecast the test series using ARIMA
    FITmodel_arima_forecast = FITmodel_arima.predict(start=split_index, end=len(x_single_user_turnover_ravel) - 1)

    # Inverse scale the ARIMA forecasted values
    FITmodel_arima_forecast = scaler.inverse_transform(FITmodel_arima_forecast.reshape(-1, 1)).reshape(-1)
     
    #SARIMA
    
    # Define the SARIMA model with seasonal difference and order
    model_sarima_monthly = SARIMAX(train_data, order=(1, 1, 1), seasonal_order=(1, 1, 1, 14))

    # Fit the model to the training data
    FITmodel_sarima_monthly = model_sarima_monthly.fit()

    # Forecast the test series
    FITmodel_sarima_monthly_forecast = FITmodel_sarima_monthly.forecast(steps=len(test_data))

    # Inverse scale the forecasted values
    FITmodel_sarima_monthly_forecast = scaler.inverse_transform(FITmodel_sarima_monthly_forecast.reshape(-1, 1)).reshape(-1)

    # Inverse scale the training data
    train_data_inverse = scaler.inverse_transform(train_data.reshape(-1, 1)).reshape(-1)

    # Inverse scale the test data
    test_data_inverse = scaler.inverse_transform(test_data.reshape(-1, 1)).reshape(-1)

    # Plot the forecast
    plt.figure(figsize=(12, 6))
    plt.plot(time_index_train, train_data_inverse, label='Train', color='blue')
    forecast_dates = pd.date_range(start=time_index_train[-1], periods=len(test_data), freq=frequency)  # Adjust as needed
    plt.plot(time_index_test, test_data_inverse, label='Test', color='orange')
    plt.plot(forecast_dates, FITmodel_sarima_monthly_forecast, label='Forecast', color='green')
    plt.xlabel('Time')
    plt.ylabel('Turnover')
    plt.legend()
    plt.title('SARIMA Forecast')
    plt.show()


    plt.figure(figsize=(12, 6))
    plt.plot(time_index_train[-48:], train_data_inverse[-48:], label='Train', color='blue', marker='o')
    plt.plot(time_index_test, test_data_inverse, label='Test', color='orange')
    plt.plot(forecast_dates, FITmodel_sarima_monthly_forecast, label='Forecast', color='green', marker='o')
    plt.xlabel('Time')
    plt.ylabel('Turnover')
    plt.legend()
    plt.title('Zoomed-in SARIMA Forecast vs Actual')
    plt.show()

    
    
     # Plot the ARIMA forecast
    plt.figure(figsize=(12, 6))
    plt.plot(time_index_train, train_data_inverse, label='Train', color='blue')
    plt.plot(time_index_test, test_data_inverse, label='Test', color='orange')
    plt.plot(forecast_dates, FITmodel_arima_forecast, label='ARIMA Forecast', color='red')
    plt.xlabel('Time')
    plt.ylabel('Turnover')
    plt.legend()
    plt.title('ARIMA Forecast')
    plt.show()
    
    
    # Calculate RMSE for ARIMA model
    rmse_arima = mean_squared_error(test_data_inverse, FITmodel_arima_forecast[:len(test_data_inverse)], squared=False)
     
    # Calculate MAE for ARIMA model
    mae_arima = np.mean(np.abs(test_data_inverse - FITmodel_arima_forecast[:len(test_data_inverse)]))
    # Display results for ARIMA model
    print(f"ARIMA Model Results (User {user_id}):")
    print(f"RMSE (ARIMA): {rmse_arima}")
    print(f"MAE (ARIMA): {mae_arima}")
    
    
    print("=" * 30)  # Separating results for different users
    
   
    
    
    
    
    #SARIMA

    # Calculate predictions for training and testing data
    train_predictions = FITmodel_sarima_monthly.predict(start=0, end=split_index - 1)
    test_predictions = FITmodel_sarima_monthly_forecast

    # Trim the predicted values to match the length of the actual values
    test_predictions_inverse_trimmed = test_predictions[:len(test_data_inverse)]

    # Calculate RMSE for training and testing
    #rmse_train = mean_squared_error(train_data_inverse, train_predictions, squared=False)
    rmse_test = mean_squared_error(test_data_inverse, test_predictions_inverse_trimmed, squared=False)
    # Calculate MAE for SARIMA model
    mae_sarima = np.mean(np.abs(test_data_inverse - test_predictions_inverse_trimmed))

    # Display results for each user
    print(f"User {user_id}:")
    print(f"SARIMAX RMSE : {rmse_test}")
    print(f"SARIMAX MAE : {mae_sarima}")
    print("=" * 30)  # Separating results for different users

    
    # Compute differences between consecutive predicted values
    differences = np.diff(test_predictions.flatten())

    # Set a threshold to identify surges
    threshold_difference = 150  

    # Identify surges based on differences and threshold
    surge_indices = np.where(differences > threshold_difference)[0]

    # Print the timestamps of the points right before a surge
    for index in surge_indices:
        if index > 0:
            surge_start_timestamp = time_index_test[index]
            print(f"PG Detected right before: {surge_start_timestamp}")

ARIMA Model Results (User 868583):
RMSE (ARIMA): 22.40285256300589
MAE (ARIMA): 22.396887663461186
==============================
User 868583:
SARIMAX RMSE : 102.5766159469979
SARIMAX MAE : 79.83388625079733
==============================
PG Detected right before: 2010-08-13 00:00:00
PG Detected right before: 2010-08-15 00:00:00
PG Detected right before: 2010-08-27 00:00:00
PG Detected right before: 2010-08-29 00:00:00
PG Detected right before: 2010-09-10 00:00:00
PG Detected right before: 2010-09-12 00:00:00
PG Detected right before: 2010-09-24 00:00:00
PG Detected right before: 2010-09-26 00:00:00
PG Detected right before: 2010-10-08 00:00:00
PG Detected right before: 2010-10-10 00:00:00
PG Detected right before: 2010-10-22 00:00:00
PG Detected right before: 2010-10-24 00:00:00

ARIMA Model Results (User 1175809):
RMSE (ARIMA): 784.7687886925128
MAE (ARIMA): 331.9608386066314
==============================
User 1175809:
SARIMAX RMSE : 777.9760114331281
SARIMAX MAE : 336.02694007952886
==============================
PG Detected right before: 2010-08-09 00:00:00
PG Detected right before: 2010-08-16 00:00:00
PG Detected right before: 2010-08-23 00:00:00
PG Detected right before: 2010-08-30 00:00:00
PG Detected right before: 2010-09-06 00:00:00
PG Detected right before: 2010-09-13 00:00:00
PG Detected right before: 2010-09-20 00:00:00
PG Detected right before: 2010-09-27 00:00:00
PG Detected right before: 2010-10-04 00:00:00
PG Detected right before: 2010-10-11 00:00:00
PG Detected right before: 2010-10-18 00:00:00
PG Detected right before: 2010-10-25 00:00:00
PG Detected right before: 2010-11-01 00:00:00

ARIMA Model Results (User 1411743):
RMSE (ARIMA): 4616.1579093677465
MAE (ARIMA): 2204.101215242136
==============================
User 1411743:
SARIMAX RMSE : 4613.733648122473
SARIMAX MAE : 2212.505857385308
==============================
PG Detected right before: 2010-08-08 00:00:00
PG Detected right before: 2010-08-12 00:00:00
PG Detected right before: 2010-08-14 00:00:00
PG Detected right before: 2010-08-15 00:00:00
PG Detected right before: 2010-08-17 00:00:00
PG Detected right before: 2010-08-21 00:00:00
PG Detected right before: 2010-08-22 00:00:00
PG Detected right before: 2010-08-26 00:00:00
PG Detected right before: 2010-08-28 00:00:00
PG Detected right before: 2010-08-29 00:00:00
PG Detected right before: 2010-08-31 00:00:00
PG Detected right before: 2010-09-04 00:00:00
PG Detected right before: 2010-09-05 00:00:00
PG Detected right before: 2010-09-09 00:00:00
PG Detected right before: 2010-09-11 00:00:00
PG Detected right before: 2010-09-12 00:00:00
PG Detected right before: 2010-09-14 00:00:00
PG Detected right before: 2010-09-18 00:00:00
PG Detected right before: 2010-09-19 00:00:00
PG Detected right before: 2010-09-23 00:00:00
PG Detected right before: 2010-09-25 00:00:00
PG Detected right before: 2010-09-26 00:00:00
PG Detected right before: 2010-09-28 00:00:00
PG Detected right before: 2010-10-02 00:00:00
PG Detected right before: 2010-10-03 00:00:00
PG Detected right before: 2010-10-07 00:00:00
PG Detected right before: 2010-10-09 00:00:00
PG Detected right before: 2010-10-10 00:00:00
PG Detected right before: 2010-10-12 00:00:00
PG Detected right before: 2010-10-16 00:00:00
PG Detected right before: 2010-10-17 00:00:00
PG Detected right before: 2010-10-21 00:00:00
PG Detected right before: 2010-10-23 00:00:00
PG Detected right before: 2010-10-24 00:00:00
PG Detected right before: 2010-10-26 00:00:00
PG Detected right before: 2010-10-30 00:00:00
PG Detected right before: 2010-10-31 00:00:00

ARIMA Model Results (User 1457496):
RMSE (ARIMA): 1978.891627932139
MAE (ARIMA): 1324.9782438380369
==============================
User 1457496:
SARIMAX RMSE : 1977.6625563070231
SARIMAX MAE : 1313.9087263358097
==============================
PG Detected right before: 2010-08-08 00:00:00
PG Detected right before: 2010-08-12 00:00:00
PG Detected right before: 2010-08-15 00:00:00
PG Detected right before: 2010-08-16 00:00:00
PG Detected right before: 2010-08-22 00:00:00
PG Detected right before: 2010-08-30 00:00:00
PG Detected right before: 2010-09-05 00:00:00
PG Detected right before: 2010-09-13 00:00:00
PG Detected right before: 2010-09-19 00:00:00
PG Detected right before: 2010-09-27 00:00:00
PG Detected right before: 2010-10-03 00:00:00
PG Detected right before: 2010-10-11 00:00:00
PG Detected right before: 2010-10-17 00:00:00
PG Detected right before: 2010-10-25 00:00:00
PG Detected right before: 2010-10-31 00:00:00

ARIMA Model Results (User 1486136):
RMSE (ARIMA): 3753.6733323996104
MAE (ARIMA): 3049.424238509906
==============================
User 1486136:
SARIMAX RMSE : 3891.2559967302295
SARIMAX MAE : 3338.0576744058267
==============================
PG Detected right before: 2010-08-12 00:00:00
PG Detected right before: 2010-08-18 00:00:00
PG Detected right before: 2010-08-21 00:00:00
PG Detected right before: 2010-08-23 00:00:00
PG Detected right before: 2010-08-26 00:00:00
PG Detected right before: 2010-08-30 00:00:00
PG Detected right before: 2010-09-06 00:00:00
PG Detected right before: 2010-09-09 00:00:00
PG Detected right before: 2010-09-13 00:00:00
PG Detected right before: 2010-09-20 00:00:00
PG Detected right before: 2010-09-23 00:00:00
PG Detected right before: 2010-09-27 00:00:00
PG Detected right before: 2010-10-04 00:00:00
PG Detected right before: 2010-10-07 00:00:00
PG Detected right before: 2010-10-11 00:00:00
PG Detected right before: 2010-10-18 00:00:00
PG Detected right before: 2010-10-21 00:00:00
PG Detected right before: 2010-10-25 00:00:00
PG Detected right before: 2010-11-01 00:00:00

ARIMA Model Results (User 1662632):
RMSE (ARIMA): 0.34753857826167966
MAE (ARIMA): 0.3459736756983271
==============================
User 1662632:
SARIMAX RMSE : 65.0293645083209
SARIMAX MAE : 53.443114830836
==============================


C:\Users\abhiv\AppData\Local\anaconda3\Lib\site-packages\statsmodels\base\model.py:604: ConvergenceWarning: Maximum Likelihood optimization failed to converge. Check mle_retvals
  warnings.warn("Maximum Likelihood optimization failed to "

ARIMA Model Results (User 1679490):
RMSE (ARIMA): 978.1364539081785
MAE (ARIMA): 636.1494596739383
==============================
User 1679490:
SARIMAX RMSE : 985.5062523969005
SARIMAX MAE : 612.500708985471
==============================
PG Detected right before: 2010-08-09 00:00:00
PG Detected right before: 2010-08-13 00:00:00
PG Detected right before: 2010-08-15 00:00:00
PG Detected right before: 2010-08-16 00:00:00
PG Detected right before: 2010-08-20 00:00:00
PG Detected right before: 2010-08-23 00:00:00
PG Detected right before: 2010-08-27 00:00:00
PG Detected right before: 2010-08-29 00:00:00
PG Detected right before: 2010-08-30 00:00:00
PG Detected right before: 2010-09-03 00:00:00
PG Detected right before: 2010-09-06 00:00:00
PG Detected right before: 2010-09-10 00:00:00
PG Detected right before: 2010-09-12 00:00:00
PG Detected right before: 2010-09-13 00:00:00
PG Detected right before: 2010-09-17 00:00:00
PG Detected right before: 2010-09-20 00:00:00
PG Detected right before: 2010-09-24 00:00:00
PG Detected right before: 2010-09-26 00:00:00
PG Detected right before: 2010-09-27 00:00:00
PG Detected right before: 2010-10-01 00:00:00
PG Detected right before: 2010-10-04 00:00:00
PG Detected right before: 2010-10-08 00:00:00
PG Detected right before: 2010-10-10 00:00:00
PG Detected right before: 2010-10-11 00:00:00
PG Detected right before: 2010-10-15 00:00:00
PG Detected right before: 2010-10-18 00:00:00
PG Detected right before: 2010-10-22 00:00:00
PG Detected right before: 2010-10-24 00:00:00
PG Detected right before: 2010-10-25 00:00:00
PG Detected right before: 2010-10-29 00:00:00
PG Detected right before: 2010-11-01 00:00:00

ARIMA Model Results (User 1790848):
RMSE (ARIMA): 0.0012906643653054297
MAE (ARIMA): 0.0012876457490887852
==============================
User 1790848:
SARIMAX RMSE : 55.19393943266972
SARIMAX MAE : 41.99327265944823
==============================

ARIMA Model Results (User 1921204):
RMSE (ARIMA): 573.9839527146021
MAE (ARIMA): 453.59999894681533
==============================
User 1921204:
SARIMAX RMSE : 564.2913244910224
SARIMAX MAE : 437.39373955004817
==============================

ARIMA Model Results (User 2070894):
RMSE (ARIMA): 1318.9026564972567
MAE (ARIMA): 939.9894409160116
==============================
User 2070894:
SARIMAX RMSE : 1338.5668002348875
SARIMAX MAE : 975.8631625342256
==============================
PG Detected right before: 2010-08-11 00:00:00
PG Detected right before: 2010-08-13 00:00:00
PG Detected right before: 2010-08-27 00:00:00
PG Detected right before: 2010-09-10 00:00:00
PG Detected right before: 2010-09-24 00:00:00
PG Detected right before: 2010-10-08 00:00:00
PG Detected right before: 2010-10-22 00:00:00


C:\Users\abhiv\AppData\Local\anaconda3\Lib\site-packages\statsmodels\base\model.py:604: ConvergenceWarning: Maximum Likelihood optimization failed to converge. Check mle_retvals
  warnings.warn("Maximum Likelihood optimization failed to "

ARIMA Model Results (User 4754125):
RMSE (ARIMA): 1771.2240395544623
MAE (ARIMA): 946.1813865979672
==============================
User 4754125:
SARIMAX RMSE : 1841.9634368882164
SARIMAX MAE : 1036.2222142263788
==============================
PG Detected right before: 2010-08-08 00:00:00
PG Detected right before: 2010-08-09 00:00:00
PG Detected right before: 2010-08-11 00:00:00
PG Detected right before: 2010-08-12 00:00:00
PG Detected right before: 2010-08-15 00:00:00
PG Detected right before: 2010-08-19 00:00:00
PG Detected right before: 2010-08-22 00:00:00
PG Detected right before: 2010-08-23 00:00:00
PG Detected right before: 2010-08-25 00:00:00
PG Detected right before: 2010-08-26 00:00:00
PG Detected right before: 2010-08-29 00:00:00
PG Detected right before: 2010-09-02 00:00:00
PG Detected right before: 2010-09-05 00:00:00
PG Detected right before: 2010-09-06 00:00:00
PG Detected right before: 2010-09-08 00:00:00
PG Detected right before: 2010-09-09 00:00:00
PG Detected right before: 2010-09-12 00:00:00
PG Detected right before: 2010-09-16 00:00:00
PG Detected right before: 2010-09-19 00:00:00
PG Detected right before: 2010-09-20 00:00:00
PG Detected right before: 2010-09-22 00:00:00
PG Detected right before: 2010-09-23 00:00:00
PG Detected right before: 2010-09-26 00:00:00
PG Detected right before: 2010-09-30 00:00:00
PG Detected right before: 2010-10-03 00:00:00
PG Detected right before: 2010-10-04 00:00:00
PG Detected right before: 2010-10-06 00:00:00
PG Detected right before: 2010-10-07 00:00:00
PG Detected right before: 2010-10-10 00:00:00
PG Detected right before: 2010-10-14 00:00:00
PG Detected right before: 2010-10-17 00:00:00
PG Detected right before: 2010-10-18 00:00:00
PG Detected right before: 2010-10-20 00:00:00
PG Detected right before: 2010-10-21 00:00:00
PG Detected right before: 2010-10-24 00:00:00
PG Detected right before: 2010-10-28 00:00:00
PG Detected right before: 2010-10-31 00:00:00
PG Detected right before: 2010-11-01 00:00:00

LSTM for non stationary users

#arimausers = 
lstmusers=is_stationary_df[(is_stationary_df['is_stationary'] == 0) | (is_stationary_df['is_stationary_hold'] == 0) | (is_stationary_df['is_stationary_NumberofBets'] == 0) |  (is_stationary_df['is_stationary_Age_until_2010'] == 0) | (is_stationary_df['is_stationary_Interventiontype_first'] == 0) ]
lstmusers=lstmusers.reset_index()
lstmusers['UserID']

0    1776178
1    3852889
2    3904422
3    4371320
4    4412550
5    4495603
6    5308271
7    6158120
8    6239380
9    6985339
Name: UserID, dtype: int64

nonstationary_single_user_data=user_data_3d['Turnover'][user_data_3d.index==lstmusers['UserID'][0]]
nonstationary_single_user_data


x_single_user_turnover_ravel=nonstationary_single_user_data.values.ravel()
x_single_user_turnover_ravel


import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from statsmodels.tsa.stattools import adfuller


# Plot the time series to visualize it
plt.plot(x_single_user_turnover_ravel)
plt.xlabel('Time')
plt.ylabel('Value')
plt.title('Time Series Data')
plt.show()

GRID SEARCH FOR BEST PARAMS

# import numpy as np
# import pandas as pd
# import matplotlib.pyplot as plt
# from sklearn.preprocessing import MinMaxScaler
# from keras.models import Sequential
# from keras.layers import Dense, LSTM
# from keras.optimizers import Adam
# from sklearn.metrics import mean_squared_error

# # Function to create and train an LSTM model
# def create_lstm_model(units, learning_rate):
#     model = Sequential()
#     model.add(LSTM(units=units, return_sequences=True, input_shape=(x_train.shape[1], 1)))
#     model.add(LSTM(units=units, return_sequences=False))
#     model.add(Dense(units=25))
#     model.add(Dense(units=1))

#     optimizer = Adam(learning_rate=learning_rate)
#     model.compile(optimizer=optimizer, loss='mean_squared_error')
    
#     return model

# # Extract statistics from the data
# row_variance = nonstationary_single_user_data.var(axis=1, skipna=True).iloc[0]
# row_mean = nonstationary_single_user_data.mean(axis=1).iloc[0]
# row_std = nonstationary_single_user_data.std(axis=1).iloc[0]

# print(f'Mean: {row_mean}')
# print(f'Standard Deviation: {row_std}')
# print(f'Variance: {row_variance}')

# # Flatten the data
# nonstationary_single_user_data_ravel = nonstationary_single_user_data.values.ravel()

# test_weeks = 3

# # Set the length of training data 
# training_data_monthly_len = len(nonstationary_single_user_data_ravel) - 7 * test_weeks

# # Scale the data
# scaler = MinMaxScaler(feature_range=(0, 1))
# scaled_data = scaler.fit_transform(nonstationary_single_user_data_ravel.reshape(-1, 1))
# train_data_monthly = scaled_data[0:int(training_data_monthly_len), :]

# # Prepare the training data
# x_train = []
# y_train = []

# for i in range(7 * test_weeks, len(train_data_monthly)):
#     x_train.append(train_data_monthly[i - 7 * test_weeks:i, 0])
#     y_train.append(train_data_monthly[i, 0])

# x_train, y_train = np.array(x_train), np.array(y_train)

# # Reshape the data
# x_train = np.reshape(x_train, (x_train.shape[0], x_train.shape[1], 1))

# # Define hyperparameters for tuning
# units_values = [64, 128, 256]
# learning_rate_values = [0.01, 0.001, 0.0001]

# best_rmse = float('inf')
# best_params = None

# # Perform grid search
# for units in units_values:
#     for learning_rate in learning_rate_values:
#         model = create_lstm_model(units, learning_rate)
        
#         # Train the model
#         model.fit(x_train, y_train, batch_size=1, epochs=3, verbose=0)

#         # Prepare the testing data
#         test_data = scaled_data[training_data_monthly_len - 7 * test_weeks:, :]
#         x_test = []

#         for i in range(7 * test_weeks, len(test_data)):
#             x_test.append(test_data[i - (7 * test_weeks):i, 0])

#         x_test = np.array(x_test)
#         x_test = np.reshape(x_test, (x_test.shape[0], x_test.shape[1], 1))

#         # Get predictions
#         predictions = model.predict(x_test)
#         predictions = scaler.inverse_transform(predictions)

#         # Get the root mean squared error (RMSE)
#         rmse = np.sqrt(mean_squared_error(predictions, nonstationary_single_user_data_ravel[training_data_monthly_len:]))
        
#         print(f"Units: {units}, Learning Rate: {learning_rate}, Test RMSE: {rmse}")

#         # Update the best parameters if RMSE is improved
#         if rmse < best_rmse:
#             best_rmse = rmse
#             best_params = {'units': units, 'learning_rate': learning_rate}

# # Print the best hyperparameters
# print(f"Best Hyperparameters: {best_params}, Best RMSE: {best_rmse}")

Saved results of LSTM Hyperparameter tuning for better reducing computation time.

# Mean: 35.68018188057653
# Standard Deviation: 122.09556811748713
# Variance: 14907.32775393194
# 1/1 [==============================] - 1s 853ms/step
# Units: 64, Learning Rate: 0.01, Test RMSE: 375.2553253207926
# 1/1 [==============================] - 1s 840ms/step
# Units: 64, Learning Rate: 0.001, Test RMSE: 231.66883181925292
# 1/1 [==============================] - 1s 907ms/step
# Units: 64, Learning Rate: 0.0001, Test RMSE: 299.90232400037746
# 1/1 [==============================] - 1s 902ms/step
# Units: 128, Learning Rate: 0.01, Test RMSE: 398.97323688410444
# 1/1 [==============================] - 1s 879ms/step
# Units: 128, Learning Rate: 0.001, Test RMSE: 241.82620950011136
# 1/1 [==============================] - 1s 854ms/step
# Units: 128, Learning Rate: 0.0001, Test RMSE: 279.9525872191042
# 1/1 [==============================] - 1s 886ms/step
# Units: 256, Learning Rate: 0.01, Test RMSE: 398.22153453188594
# 1/1 [==============================] - 1s 876ms/step
# Units: 256, Learning Rate: 0.001, Test RMSE: 289.69737162552474
# 1/1 [==============================] - 1s 879ms/step
# Units: 256, Learning Rate: 0.0001, Test RMSE: 266.59826253364014
# Best Hyperparameters: {'units': 64, 'learning_rate': 0.001}, Best RMSE: 231.66883181925292

Model Fitting of LSTM and Evaluation in Loop

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from sklearn.preprocessing import MinMaxScaler
from sklearn.metrics import mean_absolute_error
from keras.models import Sequential
from keras.layers import Dense, LSTM
from keras.optimizers import Adam
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt



for lstmuser in lstmusers['UserID']:
    


    nonstationary_single_user_data=user_data_3d['Turnover'][user_data_3d.index==lstmuser]
    nonstationary_single_user_data






    x_single_user_turnover_ravel=nonstationary_single_user_data.values.ravel()
    x_single_user_turnover_ravel




    nonstationary_single_user_data_transposed = nonstationary_single_user_data.T

    # Plot the time series with dates as columns
    plt.figure(figsize=(12, 6))
    for column in nonstationary_single_user_data_transposed.columns:
        plt.plot(nonstationary_single_user_data_transposed.index, nonstationary_single_user_data_transposed[column], label=column)

    plt.xlabel('Date')
    plt.ylabel('Turnover')
    plt.title('Time Series Data')
    plt.legend()
    plt.show()






    # Extract statistics from the data
    row_variance = nonstationary_single_user_data.var(axis=1, skipna=True).iloc[0]
    row_mean = nonstationary_single_user_data.mean(axis=1).iloc[0]
    row_std = nonstationary_single_user_data.std(axis=1).iloc[0]

    print(f'Mean: {row_mean}')
    print(f'Standard Deviation: {row_std}')
    print(f'Variance: {row_variance}')

    # Flatten the data
    nonstationary_single_user_data_ravel = nonstationary_single_user_data.values.ravel()

    test_weeks=70

    # Set the length of training data 
    training_data_monthly_len = len(nonstationary_single_user_data_ravel) - 7*test_weeks

    # Scale the data
    scaler = MinMaxScaler(feature_range=(0, 1))
    scaled_data = scaler.fit_transform(nonstationary_single_user_data_ravel.reshape(-1, 1))
    train_data_monthly = scaled_data[0:int(training_data_monthly_len), :]

    # Prepare the training data
    x_train = []
    y_train = []

    for i in range(7*test_weeks, len(train_data_monthly)):
        x_train.append(train_data_monthly[i-7*test_weeks:i, 0])
        y_train.append(train_data_monthly[i, 0])

    x_train, y_train = np.array(x_train), np.array(y_train)

    # Reshape the data
    x_train = np.reshape(x_train, (x_train.shape[0], x_train.shape[1], 1))

    # Define the best hyperparameters
    best_units = 64
    best_learning_rate = 0.001

    # Build and compile the LSTM model with the best hyperparameters
    best_model = Sequential()
    best_model.add(LSTM(units=best_units, return_sequences=True, input_shape=(x_train.shape[1], 1)))
    best_model.add(LSTM(units=best_units, return_sequences=False))
    best_model.add(Dense(units=25))
    best_model.add(Dense(units=1))

    optimizer = Adam(learning_rate=best_learning_rate)
    best_model.compile(optimizer=optimizer, loss='mean_squared_error')

    # Train the model
    best_model.fit(x_train, y_train, batch_size=1, epochs=1)  # Adjust epochs as needed

    # Prepare the testing data
    test_data = scaled_data[training_data_monthly_len - (7*test_weeks):, :]
    x_test = []

    for i in range(7*test_weeks, len(test_data)):
        x_test.append(test_data[i-(7*test_weeks):i, 0])

    x_test = np.array(x_test)
    x_test = np.reshape(x_test, (x_test.shape[0], x_test.shape[1], 1))

    # Get predictions
    predictions = best_model.predict(x_test)
    predictions = scaler.inverse_transform(predictions)

    # Get the root mean squared error (RMSE)
    rmse = np.sqrt(np.mean(((predictions - nonstationary_single_user_data_ravel[training_data_monthly_len:]) ** 2)))
    print("Test RMSE:", rmse)
    
    # Get the mean absolute error (MAE)
    mae = mean_absolute_error(nonstationary_single_user_data_ravel[training_data_monthly_len:], predictions)
    print("Test MAE:", mae)



    from datetime import datetime, timedelta
    # Generate time indices
    start_date = '2002-11-12'
    end_date = '2010-11-10'
    #test_window_lastweek='2010-11-03'


    end_date_dt = datetime.strptime(end_date, '%Y-%m-%d')

    time_index_train = pd.date_range(start=start_date,  periods=training_data_monthly_len+(1),freq=frequency)
    time_index_test = pd.date_range(start=end_date_dt - timedelta(days=(7 * (test_weeks+1)-1)), periods=len(x_test),freq=frequency)

    # Visualize the data
    train = nonstationary_single_user_data_ravel[:training_data_monthly_len+(1)]
    valid = pd.DataFrame({'x': nonstationary_single_user_data_ravel[training_data_monthly_len:]})
    valid['Predictions'] = predictions

    plt.figure(figsize=(16, 6))
    plt.title('LSTM Actual vs Predicted Turnover')
    plt.xlabel('Date', fontsize=18)
    plt.ylabel('Turnover', fontsize=18)
    plt.plot(time_index_train, train[:])
    plt.plot(time_index_test, valid['x'], linestyle='--', color='orange')
    plt.plot(time_index_test, valid['Predictions'], linestyle='--', color='green')
    plt.legend(['Train', 'Val', 'Predictions'], loc='upper right')
    plt.show()



    # Plot only the last 30 values
    last_30_values = -30

    # Plot training data
    plt.plot(time_index_train[-30:], train[last_30_values:], linestyle='-', color='blue')

    # Plot validation data
    plt.plot(time_index_test, valid['x'], linestyle='--', color='orange')  # Use x_test_idx for x-axis
    plt.plot(time_index_test, valid['Predictions'], linestyle='--', color='green')  # Use x_test_idx for x-axis

    plt.legend(['Train', 'Val', 'Predictions'], loc='upper right')

    # Rotate x-axis labels
    plt.xticks(rotation=90, ha='right')
    # Increase the number of x-axis ticks
    plt.xticks(np.arange(time_index_train[-30], time_index_test[-1], timedelta(days=14)),fontsize=7)


    plt.show()


    # Compute differences between consecutive predicted values
    differences = np.diff(predictions.flatten())

    # Set a threshold to identify surges
    threshold_difference = 0.5 

    # Identify surges based on differences and threshold
    surge_indices = np.where(differences > threshold_difference)[0]

    # Print the timestamps of the points right before a surge
    for index in surge_indices:
        if index > 0:
            surge_start_timestamp = time_index_test[index]
            print(f"PG started right before: {surge_start_timestamp}")

Mean: 386.32035549073436
Standard Deviation: 798.9267000819931
Variance: 638283.872103903
1934/1934 [==============================] - 286s 146ms/step - loss: 0.0121
16/16 [==============================] - 4s 144ms/step
Test RMSE: 699.6919454731204
Test MAE: 226.75674804189254

PG started right before: 2009-07-03 00:00:00
PG started right before: 2009-07-05 00:00:00
PG started right before: 2009-07-09 00:00:00
PG started right before: 2009-07-10 00:00:00
PG started right before: 2009-07-14 00:00:00
PG started right before: 2009-07-15 00:00:00
PG started right before: 2009-07-16 00:00:00
PG started right before: 2009-07-17 00:00:00
PG started right before: 2009-07-20 00:00:00
PG started right before: 2009-07-22 00:00:00
PG started right before: 2009-07-25 00:00:00
PG started right before: 2009-07-27 00:00:00
PG started right before: 2009-07-29 00:00:00
PG started right before: 2009-07-30 00:00:00
PG started right before: 2009-08-03 00:00:00
PG started right before: 2009-08-04 00:00:00
PG started right before: 2009-08-07 00:00:00
PG started right before: 2009-08-08 00:00:00
PG started right before: 2009-08-10 00:00:00
PG started right before: 2009-08-17 00:00:00
PG started right before: 2009-08-20 00:00:00
PG started right before: 2009-08-25 00:00:00
PG started right before: 2009-08-27 00:00:00
PG started right before: 2009-08-29 00:00:00
PG started right before: 2009-09-04 00:00:00
PG started right before: 2009-09-05 00:00:00
PG started right before: 2009-09-06 00:00:00
PG started right before: 2009-09-07 00:00:00
PG started right before: 2009-09-08 00:00:00
PG started right before: 2009-09-12 00:00:00
PG started right before: 2009-12-11 00:00:00
PG started right before: 2009-12-12 00:00:00
PG started right before: 2009-12-13 00:00:00
PG started right before: 2009-12-14 00:00:00
PG started right before: 2009-12-18 00:00:00
PG started right before: 2009-12-20 00:00:00
PG started right before: 2009-12-21 00:00:00
PG started right before: 2009-12-26 00:00:00
PG started right before: 2009-12-27 00:00:00
PG started right before: 2010-01-09 00:00:00
PG started right before: 2010-01-10 00:00:00
PG started right before: 2010-01-14 00:00:00
PG started right before: 2010-01-15 00:00:00
PG started right before: 2010-01-25 00:00:00
PG started right before: 2010-01-26 00:00:00
PG started right before: 2010-01-27 00:00:00
PG started right before: 2010-04-28 00:00:00
PG started right before: 2010-04-29 00:00:00
PG started right before: 2010-04-30 00:00:00
PG started right before: 2010-05-02 00:00:00
PG started right before: 2010-05-03 00:00:00
PG started right before: 2010-05-04 00:00:00
PG started right before: 2010-05-05 00:00:00
PG started right before: 2010-05-07 00:00:00
PG started right before: 2010-05-09 00:00:00
PG started right before: 2010-05-12 00:00:00
PG started right before: 2010-05-14 00:00:00
PG started right before: 2010-05-16 00:00:00
PG started right before: 2010-05-17 00:00:00
PG started right before: 2010-05-18 00:00:00
PG started right before: 2010-05-22 00:00:00
PG started right before: 2010-05-26 00:00:00
PG started right before: 2010-05-28 00:00:00

Mean: 365.90467207501774
Standard Deviation: 775.0542406959867
Variance: 600709.0760208324
1934/1934 [==============================] - 303s 154ms/step - loss: 6.9536e-04
16/16 [==============================] - 3s 110ms/step
Test RMSE: 1465.6427761608077
Test MAE: 844.4668944654891

PG started right before: 2009-07-03 00:00:00
PG started right before: 2009-07-04 00:00:00
PG started right before: 2009-07-06 00:00:00
PG started right before: 2009-07-07 00:00:00
PG started right before: 2009-07-08 00:00:00
PG started right before: 2009-07-10 00:00:00
PG started right before: 2009-07-11 00:00:00
PG started right before: 2009-07-16 00:00:00
PG started right before: 2009-07-22 00:00:00
PG started right before: 2009-07-23 00:00:00
PG started right before: 2009-07-24 00:00:00
PG started right before: 2009-07-25 00:00:00
PG started right before: 2009-07-26 00:00:00
PG started right before: 2009-07-27 00:00:00
PG started right before: 2009-07-28 00:00:00
PG started right before: 2009-07-29 00:00:00
PG started right before: 2009-07-30 00:00:00
PG started right before: 2009-07-31 00:00:00
PG started right before: 2009-08-01 00:00:00
PG started right before: 2009-08-02 00:00:00
PG started right before: 2009-08-03 00:00:00
PG started right before: 2009-08-05 00:00:00
PG started right before: 2009-08-06 00:00:00
PG started right before: 2009-08-07 00:00:00
PG started right before: 2009-08-08 00:00:00
PG started right before: 2009-08-09 00:00:00
PG started right before: 2009-08-26 00:00:00
PG started right before: 2009-08-27 00:00:00
PG started right before: 2009-08-28 00:00:00
PG started right before: 2009-08-30 00:00:00
PG started right before: 2009-08-31 00:00:00
PG started right before: 2009-09-17 00:00:00
PG started right before: 2009-09-18 00:00:00
PG started right before: 2009-09-21 00:00:00
PG started right before: 2009-09-22 00:00:00
PG started right before: 2009-09-23 00:00:00
PG started right before: 2009-09-24 00:00:00
PG started right before: 2009-09-25 00:00:00
PG started right before: 2009-09-26 00:00:00
PG started right before: 2009-09-27 00:00:00
PG started right before: 2009-10-11 00:00:00
PG started right before: 2009-10-14 00:00:00
PG started right before: 2009-10-17 00:00:00
PG started right before: 2009-10-20 00:00:00
PG started right before: 2009-10-21 00:00:00
PG started right before: 2009-10-22 00:00:00
PG started right before: 2009-10-23 00:00:00
PG started right before: 2009-10-24 00:00:00
PG started right before: 2009-10-27 00:00:00
PG started right before: 2009-10-31 00:00:00
PG started right before: 2009-11-01 00:00:00
PG started right before: 2009-11-02 00:00:00
PG started right before: 2009-11-03 00:00:00
PG started right before: 2009-11-07 00:00:00
PG started right before: 2009-11-15 00:00:00
PG started right before: 2009-11-17 00:00:00
PG started right before: 2009-11-18 00:00:00
PG started right before: 2009-11-19 00:00:00
PG started right before: 2009-11-23 00:00:00
PG started right before: 2009-11-24 00:00:00
PG started right before: 2009-11-28 00:00:00
PG started right before: 2009-11-30 00:00:00
PG started right before: 2009-12-01 00:00:00
PG started right before: 2009-12-02 00:00:00
PG started right before: 2009-12-03 00:00:00
PG started right before: 2009-12-07 00:00:00
PG started right before: 2009-12-08 00:00:00
PG started right before: 2009-12-09 00:00:00
PG started right before: 2009-12-10 00:00:00
PG started right before: 2009-12-11 00:00:00
PG started right before: 2009-12-12 00:00:00
PG started right before: 2009-12-13 00:00:00
PG started right before: 2009-12-15 00:00:00
PG started right before: 2009-12-16 00:00:00
PG started right before: 2009-12-19 00:00:00
PG started right before: 2009-12-20 00:00:00
PG started right before: 2009-12-30 00:00:00
PG started right before: 2010-01-05 00:00:00
PG started right before: 2010-01-06 00:00:00
PG started right before: 2010-01-14 00:00:00
PG started right before: 2010-01-15 00:00:00
PG started right before: 2010-01-16 00:00:00
PG started right before: 2010-01-17 00:00:00
PG started right before: 2010-01-18 00:00:00
PG started right before: 2010-01-19 00:00:00
PG started right before: 2010-02-02 00:00:00
PG started right before: 2010-02-03 00:00:00
PG started right before: 2010-02-04 00:00:00
PG started right before: 2010-02-05 00:00:00
PG started right before: 2010-02-09 00:00:00
PG started right before: 2010-02-19 00:00:00
PG started right before: 2010-02-20 00:00:00
PG started right before: 2010-02-21 00:00:00
PG started right before: 2010-02-23 00:00:00
PG started right before: 2010-02-24 00:00:00
PG started right before: 2010-02-25 00:00:00
PG started right before: 2010-02-26 00:00:00
PG started right before: 2010-03-08 00:00:00
PG started right before: 2010-03-16 00:00:00
PG started right before: 2010-03-17 00:00:00
PG started right before: 2010-03-22 00:00:00
PG started right before: 2010-03-23 00:00:00
PG started right before: 2010-03-24 00:00:00
PG started right before: 2010-03-25 00:00:00
PG started right before: 2010-03-26 00:00:00
PG started right before: 2010-03-30 00:00:00
PG started right before: 2010-03-31 00:00:00
PG started right before: 2010-04-01 00:00:00
PG started right before: 2010-04-02 00:00:00
PG started right before: 2010-04-03 00:00:00
PG started right before: 2010-04-04 00:00:00
PG started right before: 2010-04-05 00:00:00
PG started right before: 2010-04-06 00:00:00
PG started right before: 2010-04-07 00:00:00
PG started right before: 2010-04-09 00:00:00
PG started right before: 2010-04-13 00:00:00
PG started right before: 2010-04-14 00:00:00
PG started right before: 2010-04-15 00:00:00
PG started right before: 2010-04-16 00:00:00
PG started right before: 2010-04-17 00:00:00
PG started right before: 2010-04-18 00:00:00
PG started right before: 2010-04-19 00:00:00
PG started right before: 2010-04-20 00:00:00
PG started right before: 2010-04-21 00:00:00
PG started right before: 2010-04-29 00:00:00
PG started right before: 2010-04-30 00:00:00
PG started right before: 2010-05-01 00:00:00
PG started right before: 2010-05-02 00:00:00
PG started right before: 2010-05-03 00:00:00
PG started right before: 2010-05-04 00:00:00
PG started right before: 2010-05-05 00:00:00
PG started right before: 2010-05-06 00:00:00
PG started right before: 2010-05-11 00:00:00
PG started right before: 2010-05-18 00:00:00
PG started right before: 2010-05-19 00:00:00
PG started right before: 2010-05-24 00:00:00
PG started right before: 2010-05-26 00:00:00
PG started right before: 2010-05-28 00:00:00
PG started right before: 2010-05-31 00:00:00
PG started right before: 2010-06-01 00:00:00
PG started right before: 2010-06-02 00:00:00
PG started right before: 2010-06-03 00:00:00
PG started right before: 2010-06-04 00:00:00
PG started right before: 2010-06-05 00:00:00
PG started right before: 2010-06-07 00:00:00
PG started right before: 2010-06-08 00:00:00
PG started right before: 2010-06-09 00:00:00
PG started right before: 2010-06-13 00:00:00
PG started right before: 2010-06-14 00:00:00
PG started right before: 2010-06-15 00:00:00
PG started right before: 2010-06-16 00:00:00
PG started right before: 2010-06-19 00:00:00
PG started right before: 2010-06-21 00:00:00
PG started right before: 2010-06-22 00:00:00
PG started right before: 2010-06-23 00:00:00
PG started right before: 2010-07-06 00:00:00
PG started right before: 2010-07-10 00:00:00
PG started right before: 2010-07-12 00:00:00
PG started right before: 2010-07-13 00:00:00
PG started right before: 2010-07-22 00:00:00
PG started right before: 2010-07-27 00:00:00
PG started right before: 2010-07-28 00:00:00
PG started right before: 2010-07-29 00:00:00
PG started right before: 2010-07-30 00:00:00
PG started right before: 2010-07-31 00:00:00
PG started right before: 2010-08-01 00:00:00
PG started right before: 2010-08-02 00:00:00
PG started right before: 2010-08-08 00:00:00
PG started right before: 2010-08-11 00:00:00
PG started right before: 2010-08-13 00:00:00
PG started right before: 2010-08-14 00:00:00
PG started right before: 2010-08-15 00:00:00
PG started right before: 2010-08-17 00:00:00
PG started right before: 2010-08-31 00:00:00
PG started right before: 2010-09-05 00:00:00
PG started right before: 2010-09-06 00:00:00
PG started right before: 2010-09-07 00:00:00
PG started right before: 2010-09-08 00:00:00
PG started right before: 2010-09-09 00:00:00
PG started right before: 2010-09-10 00:00:00
PG started right before: 2010-09-11 00:00:00
PG started right before: 2010-09-12 00:00:00
PG started right before: 2010-09-13 00:00:00
PG started right before: 2010-09-14 00:00:00
PG started right before: 2010-09-15 00:00:00
PG started right before: 2010-09-24 00:00:00
PG started right before: 2010-09-25 00:00:00
PG started right before: 2010-09-27 00:00:00
PG started right before: 2010-09-28 00:00:00
PG started right before: 2010-09-29 00:00:00
PG started right before: 2010-10-02 00:00:00
PG started right before: 2010-10-11 00:00:00
PG started right before: 2010-10-12 00:00:00
PG started right before: 2010-10-13 00:00:00
PG started right before: 2010-10-14 00:00:00
PG started right before: 2010-10-15 00:00:00
PG started right before: 2010-10-16 00:00:00
PG started right before: 2010-10-21 00:00:00
PG started right before: 2010-10-27 00:00:00

Mean: 297.5882071771934
Standard Deviation: 912.9051402201292
Variance: 833395.7950403338
1934/1934 [==============================] - 258s 132ms/step - loss: 0.0041
16/16 [==============================] - 3s 114ms/step
Test RMSE: 1438.35631447229
Test MAE: 607.3983705502761

PG started right before: 2009-08-05 00:00:00
PG started right before: 2009-08-06 00:00:00
PG started right before: 2009-08-07 00:00:00
PG started right before: 2009-08-09 00:00:00
PG started right before: 2009-08-10 00:00:00
PG started right before: 2009-08-11 00:00:00
PG started right before: 2009-08-12 00:00:00
PG started right before: 2009-08-14 00:00:00
PG started right before: 2009-08-16 00:00:00
PG started right before: 2009-08-17 00:00:00
PG started right before: 2009-08-18 00:00:00
PG started right before: 2009-08-19 00:00:00
PG started right before: 2009-08-23 00:00:00
PG started right before: 2009-08-25 00:00:00
PG started right before: 2009-08-27 00:00:00
PG started right before: 2009-08-31 00:00:00
PG started right before: 2009-09-04 00:00:00
PG started right before: 2009-09-07 00:00:00
PG started right before: 2009-09-09 00:00:00
PG started right before: 2009-09-12 00:00:00
PG started right before: 2009-09-14 00:00:00
PG started right before: 2009-09-15 00:00:00
PG started right before: 2009-09-17 00:00:00
PG started right before: 2009-09-18 00:00:00
PG started right before: 2009-10-08 00:00:00
PG started right before: 2009-10-13 00:00:00
PG started right before: 2009-10-14 00:00:00
PG started right before: 2009-10-18 00:00:00
PG started right before: 2009-10-20 00:00:00
PG started right before: 2009-10-21 00:00:00
PG started right before: 2009-10-23 00:00:00
PG started right before: 2009-10-29 00:00:00
PG started right before: 2009-11-02 00:00:00
PG started right before: 2009-11-03 00:00:00
PG started right before: 2009-11-07 00:00:00
PG started right before: 2009-11-12 00:00:00
PG started right before: 2009-11-13 00:00:00
PG started right before: 2009-11-17 00:00:00
PG started right before: 2009-11-25 00:00:00
PG started right before: 2009-11-29 00:00:00
PG started right before: 2009-11-30 00:00:00
PG started right before: 2009-12-01 00:00:00
PG started right before: 2009-12-09 00:00:00
PG started right before: 2009-12-10 00:00:00
PG started right before: 2009-12-13 00:00:00
PG started right before: 2009-12-14 00:00:00
PG started right before: 2009-12-15 00:00:00
PG started right before: 2009-12-19 00:00:00
PG started right before: 2009-12-20 00:00:00
PG started right before: 2010-01-02 00:00:00
PG started right before: 2010-01-03 00:00:00
PG started right before: 2010-01-05 00:00:00
PG started right before: 2010-01-08 00:00:00
PG started right before: 2010-01-11 00:00:00
PG started right before: 2010-01-12 00:00:00
PG started right before: 2010-01-14 00:00:00
PG started right before: 2010-01-15 00:00:00
PG started right before: 2010-01-16 00:00:00
PG started right before: 2010-01-20 00:00:00
PG started right before: 2010-01-26 00:00:00
PG started right before: 2010-02-01 00:00:00
PG started right before: 2010-02-02 00:00:00
PG started right before: 2010-02-03 00:00:00
PG started right before: 2010-02-05 00:00:00
PG started right before: 2010-02-06 00:00:00
PG started right before: 2010-02-09 00:00:00
PG started right before: 2010-02-10 00:00:00
PG started right before: 2010-02-11 00:00:00
PG started right before: 2010-02-18 00:00:00
PG started right before: 2010-02-20 00:00:00
PG started right before: 2010-02-23 00:00:00
PG started right before: 2010-02-28 00:00:00
PG started right before: 2010-03-02 00:00:00
PG started right before: 2010-03-03 00:00:00
PG started right before: 2010-03-08 00:00:00
PG started right before: 2010-03-09 00:00:00
PG started right before: 2010-03-12 00:00:00
PG started right before: 2010-03-13 00:00:00
PG started right before: 2010-03-17 00:00:00
PG started right before: 2010-03-20 00:00:00
PG started right before: 2010-03-25 00:00:00
PG started right before: 2010-03-26 00:00:00
PG started right before: 2010-03-28 00:00:00
PG started right before: 2010-04-06 00:00:00
PG started right before: 2010-04-07 00:00:00
PG started right before: 2010-04-14 00:00:00
PG started right before: 2010-04-15 00:00:00
PG started right before: 2010-04-16 00:00:00
PG started right before: 2010-04-17 00:00:00
PG started right before: 2010-04-22 00:00:00
PG started right before: 2010-04-26 00:00:00
PG started right before: 2010-04-27 00:00:00
PG started right before: 2010-05-01 00:00:00
PG started right before: 2010-05-05 00:00:00
PG started right before: 2010-05-08 00:00:00
PG started right before: 2010-05-17 00:00:00
PG started right before: 2010-05-18 00:00:00
PG started right before: 2010-05-22 00:00:00
PG started right before: 2010-05-24 00:00:00
PG started right before: 2010-05-27 00:00:00
PG started right before: 2010-05-30 00:00:00
PG started right before: 2010-06-01 00:00:00
PG started right before: 2010-06-02 00:00:00
PG started right before: 2010-06-05 00:00:00
PG started right before: 2010-06-09 00:00:00
PG started right before: 2010-06-12 00:00:00
PG started right before: 2010-06-15 00:00:00
PG started right before: 2010-06-20 00:00:00
PG started right before: 2010-07-10 00:00:00
PG started right before: 2010-07-14 00:00:00
PG started right before: 2010-07-20 00:00:00
PG started right before: 2010-07-23 00:00:00
PG started right before: 2010-07-27 00:00:00
PG started right before: 2010-07-28 00:00:00
PG started right before: 2010-07-30 00:00:00
PG started right before: 2010-07-31 00:00:00
PG started right before: 2010-08-01 00:00:00
PG started right before: 2010-08-03 00:00:00
PG started right before: 2010-08-06 00:00:00
PG started right before: 2010-08-08 00:00:00
PG started right before: 2010-08-09 00:00:00
PG started right before: 2010-08-20 00:00:00
PG started right before: 2010-08-30 00:00:00
PG started right before: 2010-08-31 00:00:00
PG started right before: 2010-09-02 00:00:00
PG started right before: 2010-09-07 00:00:00
PG started right before: 2010-09-09 00:00:00
PG started right before: 2010-09-16 00:00:00
PG started right before: 2010-09-19 00:00:00
PG started right before: 2010-09-22 00:00:00
PG started right before: 2010-09-24 00:00:00
PG started right before: 2010-09-28 00:00:00
PG started right before: 2010-09-29 00:00:00
PG started right before: 2010-09-30 00:00:00
PG started right before: 2010-10-01 00:00:00
PG started right before: 2010-10-05 00:00:00
PG started right before: 2010-10-11 00:00:00
PG started right before: 2010-10-14 00:00:00
PG started right before: 2010-11-02 00:00:00

Mean: 276.44176282570834
Standard Deviation: 949.0362785545946
Variance: 900669.858012754
1934/1934 [==============================] - 326s 167ms/step - loss: 0.0016
16/16 [==============================] - 5s 211ms/step
Test RMSE: 1904.0475907483478
Test MAE: 938.5964909608357

PG started right before: 2009-07-03 00:00:00
PG started right before: 2009-07-04 00:00:00
PG started right before: 2009-07-08 00:00:00
PG started right before: 2009-07-09 00:00:00
PG started right before: 2009-07-10 00:00:00
PG started right before: 2009-07-14 00:00:00
PG started right before: 2009-07-17 00:00:00
PG started right before: 2009-07-18 00:00:00
PG started right before: 2009-07-20 00:00:00
PG started right before: 2009-07-21 00:00:00
PG started right before: 2009-07-22 00:00:00
PG started right before: 2009-08-05 00:00:00
PG started right before: 2009-08-15 00:00:00
PG started right before: 2009-08-16 00:00:00
PG started right before: 2009-08-17 00:00:00
PG started right before: 2009-08-22 00:00:00
PG started right before: 2009-08-23 00:00:00
PG started right before: 2009-08-24 00:00:00
PG started right before: 2009-08-25 00:00:00
PG started right before: 2009-08-27 00:00:00
PG started right before: 2009-08-30 00:00:00
PG started right before: 2009-08-31 00:00:00
PG started right before: 2009-09-02 00:00:00
PG started right before: 2009-09-03 00:00:00
PG started right before: 2009-09-08 00:00:00
PG started right before: 2009-09-11 00:00:00
PG started right before: 2009-09-12 00:00:00
PG started right before: 2009-09-13 00:00:00
PG started right before: 2009-09-14 00:00:00
PG started right before: 2009-09-17 00:00:00
PG started right before: 2009-09-21 00:00:00
PG started right before: 2009-09-22 00:00:00
PG started right before: 2009-09-27 00:00:00
PG started right before: 2009-09-28 00:00:00
PG started right before: 2009-09-29 00:00:00
PG started right before: 2009-09-30 00:00:00
PG started right before: 2009-10-02 00:00:00
PG started right before: 2009-10-03 00:00:00
PG started right before: 2009-10-04 00:00:00
PG started right before: 2009-10-13 00:00:00
PG started right before: 2009-10-14 00:00:00
PG started right before: 2009-10-20 00:00:00
PG started right before: 2009-10-29 00:00:00
PG started right before: 2009-10-30 00:00:00
PG started right before: 2009-11-01 00:00:00
PG started right before: 2009-11-03 00:00:00
PG started right before: 2009-11-05 00:00:00
PG started right before: 2009-11-06 00:00:00
PG started right before: 2009-11-07 00:00:00
PG started right before: 2009-11-08 00:00:00
PG started right before: 2009-11-09 00:00:00
PG started right before: 2009-11-11 00:00:00
PG started right before: 2009-11-12 00:00:00
PG started right before: 2009-11-21 00:00:00
PG started right before: 2009-11-24 00:00:00
PG started right before: 2009-12-03 00:00:00
PG started right before: 2009-12-04 00:00:00
PG started right before: 2009-12-10 00:00:00
PG started right before: 2009-12-11 00:00:00
PG started right before: 2009-12-12 00:00:00
PG started right before: 2009-12-19 00:00:00
PG started right before: 2009-12-22 00:00:00
PG started right before: 2009-12-26 00:00:00
PG started right before: 2009-12-27 00:00:00
PG started right before: 2010-01-03 00:00:00
PG started right before: 2010-01-04 00:00:00
PG started right before: 2010-01-06 00:00:00
PG started right before: 2010-01-08 00:00:00
PG started right before: 2010-01-09 00:00:00
PG started right before: 2010-01-12 00:00:00
PG started right before: 2010-01-13 00:00:00
PG started right before: 2010-01-14 00:00:00
PG started right before: 2010-01-18 00:00:00
PG started right before: 2010-01-19 00:00:00
PG started right before: 2010-01-20 00:00:00
PG started right before: 2010-01-21 00:00:00
PG started right before: 2010-01-22 00:00:00
PG started right before: 2010-01-23 00:00:00
PG started right before: 2010-01-30 00:00:00
PG started right before: 2010-02-01 00:00:00
PG started right before: 2010-02-03 00:00:00
PG started right before: 2010-02-04 00:00:00
PG started right before: 2010-02-14 00:00:00
PG started right before: 2010-02-21 00:00:00
PG started right before: 2010-02-24 00:00:00
PG started right before: 2010-02-25 00:00:00
PG started right before: 2010-03-03 00:00:00
PG started right before: 2010-03-07 00:00:00
PG started right before: 2010-03-08 00:00:00
PG started right before: 2010-03-09 00:00:00
PG started right before: 2010-03-19 00:00:00
PG started right before: 2010-03-20 00:00:00
PG started right before: 2010-03-21 00:00:00
PG started right before: 2010-03-23 00:00:00
PG started right before: 2010-03-24 00:00:00
PG started right before: 2010-03-26 00:00:00
PG started right before: 2010-03-27 00:00:00
PG started right before: 2010-03-28 00:00:00
PG started right before: 2010-03-29 00:00:00
PG started right before: 2010-04-01 00:00:00
PG started right before: 2010-04-02 00:00:00
PG started right before: 2010-04-03 00:00:00
PG started right before: 2010-04-13 00:00:00
PG started right before: 2010-04-14 00:00:00
PG started right before: 2010-04-15 00:00:00
PG started right before: 2010-04-17 00:00:00
PG started right before: 2010-04-23 00:00:00
PG started right before: 2010-04-24 00:00:00
PG started right before: 2010-05-20 00:00:00
PG started right before: 2010-05-21 00:00:00
PG started right before: 2010-05-22 00:00:00
PG started right before: 2010-06-09 00:00:00
PG started right before: 2010-06-15 00:00:00
PG started right before: 2010-06-16 00:00:00
PG started right before: 2010-06-18 00:00:00
PG started right before: 2010-06-23 00:00:00
PG started right before: 2010-06-24 00:00:00
PG started right before: 2010-06-28 00:00:00
PG started right before: 2010-06-30 00:00:00
PG started right before: 2010-07-05 00:00:00
PG started right before: 2010-07-06 00:00:00
PG started right before: 2010-07-07 00:00:00
PG started right before: 2010-07-24 00:00:00
PG started right before: 2010-07-27 00:00:00
PG started right before: 2010-07-28 00:00:00
PG started right before: 2010-07-29 00:00:00
PG started right before: 2010-07-30 00:00:00
PG started right before: 2010-07-31 00:00:00
PG started right before: 2010-08-01 00:00:00
PG started right before: 2010-08-08 00:00:00
PG started right before: 2010-08-09 00:00:00
PG started right before: 2010-08-20 00:00:00
PG started right before: 2010-08-21 00:00:00
PG started right before: 2010-08-22 00:00:00
PG started right before: 2010-08-23 00:00:00
PG started right before: 2010-08-25 00:00:00
PG started right before: 2010-08-30 00:00:00
PG started right before: 2010-09-02 00:00:00
PG started right before: 2010-09-03 00:00:00
PG started right before: 2010-09-10 00:00:00
PG started right before: 2010-09-11 00:00:00
PG started right before: 2010-09-12 00:00:00
PG started right before: 2010-09-20 00:00:00
PG started right before: 2010-09-21 00:00:00
PG started right before: 2010-09-24 00:00:00
PG started right before: 2010-09-25 00:00:00
PG started right before: 2010-09-28 00:00:00
PG started right before: 2010-09-29 00:00:00
PG started right before: 2010-09-30 00:00:00
PG started right before: 2010-10-05 00:00:00
PG started right before: 2010-10-06 00:00:00
PG started right before: 2010-10-15 00:00:00
PG started right before: 2010-10-17 00:00:00
PG started right before: 2010-10-18 00:00:00
PG started right before: 2010-10-21 00:00:00
PG started right before: 2010-10-23 00:00:00
PG started right before: 2010-10-25 00:00:00

Mean: 239.93016815374057
Standard Deviation: 1034.4351019416395
Variance: 1070055.98012901
1934/1934 [==============================] - 397s 202ms/step - loss: 0.0034
16/16 [==============================] - 5s 199ms/step
Test RMSE: 817.9082970425846
Test MAE: 435.18406274663187

PG started right before: 2009-07-16 00:00:00
PG started right before: 2009-08-13 00:00:00
PG started right before: 2009-08-24 00:00:00
PG started right before: 2009-08-25 00:00:00
PG started right before: 2009-08-26 00:00:00
PG started right before: 2009-09-05 00:00:00
PG started right before: 2009-09-08 00:00:00
PG started right before: 2009-09-22 00:00:00
PG started right before: 2009-09-23 00:00:00
PG started right before: 2009-09-25 00:00:00
PG started right before: 2009-09-26 00:00:00
PG started right before: 2009-10-22 00:00:00
PG started right before: 2009-10-23 00:00:00
PG started right before: 2009-10-25 00:00:00
PG started right before: 2009-10-26 00:00:00
PG started right before: 2009-11-17 00:00:00
PG started right before: 2009-11-20 00:00:00
PG started right before: 2009-11-21 00:00:00
PG started right before: 2009-11-22 00:00:00
PG started right before: 2009-11-24 00:00:00
PG started right before: 2009-11-25 00:00:00
PG started right before: 2009-12-15 00:00:00
PG started right before: 2009-12-17 00:00:00
PG started right before: 2009-12-18 00:00:00
PG started right before: 2009-12-21 00:00:00
PG started right before: 2010-01-20 00:00:00
PG started right before: 2010-01-21 00:00:00
PG started right before: 2010-01-22 00:00:00
PG started right before: 2010-01-28 00:00:00
PG started right before: 2010-02-06 00:00:00
PG started right before: 2010-02-13 00:00:00
PG started right before: 2010-02-14 00:00:00
PG started right before: 2010-02-17 00:00:00
PG started right before: 2010-02-18 00:00:00
PG started right before: 2010-02-22 00:00:00
PG started right before: 2010-02-24 00:00:00
PG started right before: 2010-02-25 00:00:00
PG started right before: 2010-03-19 00:00:00
PG started right before: 2010-03-20 00:00:00
PG started right before: 2010-03-21 00:00:00
PG started right before: 2010-03-27 00:00:00
PG started right before: 2010-03-28 00:00:00
PG started right before: 2010-04-02 00:00:00
PG started right before: 2010-04-03 00:00:00
PG started right before: 2010-04-21 00:00:00
PG started right before: 2010-04-25 00:00:00
PG started right before: 2010-04-26 00:00:00
PG started right before: 2010-04-27 00:00:00
PG started right before: 2010-04-28 00:00:00
PG started right before: 2010-05-03 00:00:00
PG started right before: 2010-05-04 00:00:00
PG started right before: 2010-05-07 00:00:00
PG started right before: 2010-05-08 00:00:00

Mean: 445.9454516561502
Standard Deviation: 1455.7747813704814
Variance: 2119280.2140742727
1934/1934 [==============================] - 381s 194ms/step - loss: 0.0056
16/16 [==============================] - 5s 201ms/step
Test RMSE: 452.7626105463848
Test MAE: 116.34511226989632

PG started right before: 2009-07-03 00:00:00
PG started right before: 2009-07-04 00:00:00
PG started right before: 2009-07-05 00:00:00
PG started right before: 2009-07-17 00:00:00
PG started right before: 2009-07-18 00:00:00
PG started right before: 2009-07-22 00:00:00
PG started right before: 2009-07-23 00:00:00
PG started right before: 2009-07-24 00:00:00
PG started right before: 2009-07-25 00:00:00
PG started right before: 2009-08-03 00:00:00
PG started right before: 2009-08-04 00:00:00
PG started right before: 2009-08-11 00:00:00
PG started right before: 2009-09-18 00:00:00
PG started right before: 2009-09-19 00:00:00
PG started right before: 2009-10-09 00:00:00
PG started right before: 2009-11-27 00:00:00
PG started right before: 2009-11-28 00:00:00

Mean: 555.0427506863418
Standard Deviation: 1780.5543780564858
Variance: 3170373.893216119
1934/1934 [==============================] - 301s 152ms/step - loss: 0.0038
16/16 [==============================] - 3s 128ms/step
Test RMSE: 2993.7127535504333
Test MAE: 1485.7203724097428

PG started right before: 2009-07-04 00:00:00
PG started right before: 2009-07-08 00:00:00
PG started right before: 2009-07-09 00:00:00
PG started right before: 2009-07-10 00:00:00
PG started right before: 2009-07-11 00:00:00
PG started right before: 2009-07-12 00:00:00
PG started right before: 2009-07-18 00:00:00
PG started right before: 2009-07-19 00:00:00
PG started right before: 2009-07-23 00:00:00
PG started right before: 2009-07-24 00:00:00
PG started right before: 2009-07-25 00:00:00
PG started right before: 2009-07-26 00:00:00
PG started right before: 2009-08-01 00:00:00
PG started right before: 2009-08-04 00:00:00
PG started right before: 2009-08-05 00:00:00
PG started right before: 2009-08-08 00:00:00
PG started right before: 2009-08-09 00:00:00
PG started right before: 2009-08-11 00:00:00
PG started right before: 2009-08-20 00:00:00
PG started right before: 2009-08-21 00:00:00
PG started right before: 2009-08-22 00:00:00
PG started right before: 2009-08-27 00:00:00
PG started right before: 2009-08-28 00:00:00
PG started right before: 2009-08-29 00:00:00
PG started right before: 2009-09-02 00:00:00
PG started right before: 2009-09-04 00:00:00
PG started right before: 2009-09-05 00:00:00
PG started right before: 2009-09-06 00:00:00
PG started right before: 2009-09-08 00:00:00
PG started right before: 2009-09-09 00:00:00
PG started right before: 2009-09-10 00:00:00
PG started right before: 2009-09-11 00:00:00
PG started right before: 2009-09-22 00:00:00
PG started right before: 2009-09-23 00:00:00
PG started right before: 2009-09-24 00:00:00
PG started right before: 2009-09-25 00:00:00
PG started right before: 2009-09-28 00:00:00
PG started right before: 2009-09-29 00:00:00
PG started right before: 2009-10-01 00:00:00
PG started right before: 2009-10-03 00:00:00
PG started right before: 2009-10-06 00:00:00
PG started right before: 2009-10-08 00:00:00
PG started right before: 2009-10-09 00:00:00
PG started right before: 2009-10-10 00:00:00
PG started right before: 2009-10-11 00:00:00
PG started right before: 2009-10-15 00:00:00
PG started right before: 2009-10-16 00:00:00
PG started right before: 2009-10-19 00:00:00
PG started right before: 2009-10-20 00:00:00
PG started right before: 2009-10-21 00:00:00
PG started right before: 2009-10-22 00:00:00
PG started right before: 2009-10-23 00:00:00
PG started right before: 2009-10-25 00:00:00
PG started right before: 2009-10-30 00:00:00
PG started right before: 2009-11-01 00:00:00
PG started right before: 2009-11-05 00:00:00
PG started right before: 2009-11-06 00:00:00
PG started right before: 2009-11-10 00:00:00
PG started right before: 2009-11-19 00:00:00
PG started right before: 2009-11-20 00:00:00
PG started right before: 2009-11-21 00:00:00
PG started right before: 2009-11-22 00:00:00
PG started right before: 2009-11-24 00:00:00
PG started right before: 2009-11-26 00:00:00
PG started right before: 2009-11-28 00:00:00
PG started right before: 2009-12-02 00:00:00
PG started right before: 2009-12-03 00:00:00
PG started right before: 2009-12-04 00:00:00
PG started right before: 2009-12-05 00:00:00
PG started right before: 2009-12-08 00:00:00
PG started right before: 2009-12-09 00:00:00
PG started right before: 2009-12-14 00:00:00
PG started right before: 2009-12-15 00:00:00
PG started right before: 2009-12-19 00:00:00
PG started right before: 2009-12-22 00:00:00
PG started right before: 2009-12-27 00:00:00
PG started right before: 2009-12-28 00:00:00
PG started right before: 2009-12-29 00:00:00
PG started right before: 2010-01-01 00:00:00
PG started right before: 2010-01-10 00:00:00
PG started right before: 2010-01-12 00:00:00
PG started right before: 2010-01-13 00:00:00
PG started right before: 2010-01-21 00:00:00
PG started right before: 2010-01-22 00:00:00
PG started right before: 2010-01-23 00:00:00
PG started right before: 2010-01-24 00:00:00
PG started right before: 2010-01-27 00:00:00
PG started right before: 2010-01-28 00:00:00
PG started right before: 2010-02-23 00:00:00
PG started right before: 2010-03-01 00:00:00
PG started right before: 2010-03-02 00:00:00
PG started right before: 2010-03-05 00:00:00
PG started right before: 2010-03-06 00:00:00
PG started right before: 2010-03-07 00:00:00
PG started right before: 2010-03-08 00:00:00
PG started right before: 2010-03-09 00:00:00
PG started right before: 2010-03-11 00:00:00
PG started right before: 2010-03-12 00:00:00
PG started right before: 2010-03-23 00:00:00
PG started right before: 2010-03-24 00:00:00
PG started right before: 2010-03-26 00:00:00
PG started right before: 2010-04-04 00:00:00
PG started right before: 2010-04-05 00:00:00
PG started right before: 2010-04-08 00:00:00
PG started right before: 2010-04-20 00:00:00
PG started right before: 2010-04-22 00:00:00
PG started right before: 2010-04-23 00:00:00
PG started right before: 2010-04-24 00:00:00
PG started right before: 2010-04-29 00:00:00
PG started right before: 2010-05-03 00:00:00
PG started right before: 2010-05-04 00:00:00
PG started right before: 2010-05-07 00:00:00
PG started right before: 2010-05-08 00:00:00
PG started right before: 2010-05-19 00:00:00
PG started right before: 2010-05-20 00:00:00
PG started right before: 2010-05-21 00:00:00
PG started right before: 2010-05-26 00:00:00
PG started right before: 2010-05-27 00:00:00
PG started right before: 2010-05-28 00:00:00
PG started right before: 2010-05-29 00:00:00
PG started right before: 2010-06-01 00:00:00
PG started right before: 2010-06-10 00:00:00
PG started right before: 2010-06-11 00:00:00
PG started right before: 2010-06-12 00:00:00
PG started right before: 2010-06-20 00:00:00
PG started right before: 2010-06-21 00:00:00
PG started right before: 2010-06-22 00:00:00
PG started right before: 2010-06-23 00:00:00
PG started right before: 2010-06-24 00:00:00
PG started right before: 2010-06-26 00:00:00
PG started right before: 2010-06-30 00:00:00
PG started right before: 2010-07-10 00:00:00
PG started right before: 2010-07-11 00:00:00
PG started right before: 2010-07-24 00:00:00
PG started right before: 2010-07-27 00:00:00
PG started right before: 2010-07-28 00:00:00
PG started right before: 2010-08-01 00:00:00
PG started right before: 2010-08-02 00:00:00
PG started right before: 2010-08-03 00:00:00
PG started right before: 2010-08-20 00:00:00
PG started right before: 2010-08-21 00:00:00
PG started right before: 2010-08-22 00:00:00
PG started right before: 2010-08-23 00:00:00
PG started right before: 2010-08-24 00:00:00
PG started right before: 2010-08-26 00:00:00
PG started right before: 2010-09-01 00:00:00
PG started right before: 2010-09-02 00:00:00
PG started right before: 2010-09-03 00:00:00
PG started right before: 2010-09-04 00:00:00
PG started right before: 2010-09-09 00:00:00
PG started right before: 2010-09-10 00:00:00
PG started right before: 2010-09-22 00:00:00
PG started right before: 2010-09-24 00:00:00
PG started right before: 2010-09-30 00:00:00
PG started right before: 2010-10-01 00:00:00
PG started right before: 2010-10-02 00:00:00
PG started right before: 2010-10-05 00:00:00
PG started right before: 2010-10-07 00:00:00
PG started right before: 2010-10-08 00:00:00
PG started right before: 2010-10-09 00:00:00
PG started right before: 2010-10-13 00:00:00
PG started right before: 2010-10-19 00:00:00
PG started right before: 2010-10-20 00:00:00
PG started right before: 2010-10-23 00:00:00
PG started right before: 2010-10-24 00:00:00
PG started right before: 2010-10-29 00:00:00
PG started right before: 2010-10-30 00:00:00

Mean: 238.77552505147563
Standard Deviation: 980.4562732027541
Variance: 961294.5036626337
1934/1934 [==============================] - 261s 133ms/step - loss: 0.0010
16/16 [==============================] - 3s 139ms/step
Test RMSE: 2064.926045109493
Test MAE: 878.5353481787856

PG started right before: 2009-07-06 00:00:00
PG started right before: 2009-07-07 00:00:00
PG started right before: 2009-07-08 00:00:00
PG started right before: 2009-07-09 00:00:00
PG started right before: 2009-07-10 00:00:00
PG started right before: 2009-07-11 00:00:00
PG started right before: 2009-07-12 00:00:00
PG started right before: 2009-07-18 00:00:00
PG started right before: 2009-07-22 00:00:00
PG started right before: 2009-07-23 00:00:00
PG started right before: 2009-07-25 00:00:00
PG started right before: 2009-07-26 00:00:00
PG started right before: 2009-07-28 00:00:00
PG started right before: 2009-07-30 00:00:00
PG started right before: 2009-07-31 00:00:00
PG started right before: 2009-08-01 00:00:00
PG started right before: 2009-08-07 00:00:00
PG started right before: 2009-08-11 00:00:00
PG started right before: 2009-08-12 00:00:00
PG started right before: 2009-08-13 00:00:00
PG started right before: 2009-08-20 00:00:00
PG started right before: 2009-08-21 00:00:00
PG started right before: 2009-08-22 00:00:00
PG started right before: 2009-08-23 00:00:00
PG started right before: 2009-08-28 00:00:00
PG started right before: 2009-09-04 00:00:00
PG started right before: 2009-09-05 00:00:00
PG started right before: 2009-09-06 00:00:00
PG started right before: 2009-09-08 00:00:00
PG started right before: 2009-09-09 00:00:00
PG started right before: 2009-09-10 00:00:00
PG started right before: 2009-09-13 00:00:00
PG started right before: 2009-09-20 00:00:00
PG started right before: 2009-09-21 00:00:00
PG started right before: 2009-09-22 00:00:00
PG started right before: 2009-09-24 00:00:00
PG started right before: 2009-09-25 00:00:00
PG started right before: 2009-09-26 00:00:00
PG started right before: 2009-09-28 00:00:00
PG started right before: 2009-10-01 00:00:00
PG started right before: 2009-10-02 00:00:00
PG started right before: 2009-10-04 00:00:00
PG started right before: 2009-10-08 00:00:00
PG started right before: 2009-10-09 00:00:00
PG started right before: 2009-10-10 00:00:00
PG started right before: 2009-10-11 00:00:00
PG started right before: 2009-10-13 00:00:00
PG started right before: 2009-10-19 00:00:00
PG started right before: 2009-10-20 00:00:00
PG started right before: 2009-10-21 00:00:00
PG started right before: 2009-10-22 00:00:00
PG started right before: 2009-10-23 00:00:00
PG started right before: 2009-10-27 00:00:00
PG started right before: 2009-10-28 00:00:00
PG started right before: 2009-10-29 00:00:00
PG started right before: 2009-10-30 00:00:00
PG started right before: 2009-10-31 00:00:00
PG started right before: 2009-11-01 00:00:00
PG started right before: 2009-11-05 00:00:00
PG started right before: 2009-11-06 00:00:00
PG started right before: 2009-11-08 00:00:00
PG started right before: 2009-11-10 00:00:00
PG started right before: 2009-11-11 00:00:00
PG started right before: 2009-11-16 00:00:00
PG started right before: 2009-11-18 00:00:00
PG started right before: 2009-11-19 00:00:00
PG started right before: 2009-11-24 00:00:00
PG started right before: 2009-11-25 00:00:00
PG started right before: 2009-11-26 00:00:00
PG started right before: 2009-11-28 00:00:00
PG started right before: 2009-12-03 00:00:00
PG started right before: 2009-12-05 00:00:00
PG started right before: 2009-12-06 00:00:00
PG started right before: 2009-12-07 00:00:00
PG started right before: 2009-12-08 00:00:00
PG started right before: 2009-12-09 00:00:00
PG started right before: 2009-12-10 00:00:00
PG started right before: 2009-12-11 00:00:00
PG started right before: 2009-12-12 00:00:00
PG started right before: 2009-12-13 00:00:00
PG started right before: 2009-12-19 00:00:00
PG started right before: 2009-12-20 00:00:00
PG started right before: 2009-12-21 00:00:00
PG started right before: 2009-12-23 00:00:00
PG started right before: 2009-12-25 00:00:00
PG started right before: 2009-12-26 00:00:00
PG started right before: 2009-12-27 00:00:00
PG started right before: 2009-12-31 00:00:00
PG started right before: 2010-01-01 00:00:00
PG started right before: 2010-01-02 00:00:00
PG started right before: 2010-01-03 00:00:00
PG started right before: 2010-01-08 00:00:00
PG started right before: 2010-01-09 00:00:00
PG started right before: 2010-01-12 00:00:00
PG started right before: 2010-01-15 00:00:00
PG started right before: 2010-01-21 00:00:00
PG started right before: 2010-01-22 00:00:00
PG started right before: 2010-01-23 00:00:00
PG started right before: 2010-01-24 00:00:00
PG started right before: 2010-01-30 00:00:00
PG started right before: 2010-01-31 00:00:00
PG started right before: 2010-02-02 00:00:00
PG started right before: 2010-02-03 00:00:00
PG started right before: 2010-02-05 00:00:00
PG started right before: 2010-02-06 00:00:00
PG started right before: 2010-02-07 00:00:00
PG started right before: 2010-02-13 00:00:00
PG started right before: 2010-02-14 00:00:00
PG started right before: 2010-02-15 00:00:00
PG started right before: 2010-02-16 00:00:00
PG started right before: 2010-02-17 00:00:00
PG started right before: 2010-02-18 00:00:00
PG started right before: 2010-02-20 00:00:00
PG started right before: 2010-02-21 00:00:00
PG started right before: 2010-02-24 00:00:00
PG started right before: 2010-02-27 00:00:00
PG started right before: 2010-02-28 00:00:00
PG started right before: 2010-03-06 00:00:00
PG started right before: 2010-03-07 00:00:00
PG started right before: 2010-03-25 00:00:00
PG started right before: 2010-03-26 00:00:00
PG started right before: 2010-03-31 00:00:00
PG started right before: 2010-04-01 00:00:00
PG started right before: 2010-04-02 00:00:00
PG started right before: 2010-04-07 00:00:00
PG started right before: 2010-04-08 00:00:00
PG started right before: 2010-04-10 00:00:00
PG started right before: 2010-04-13 00:00:00
PG started right before: 2010-04-23 00:00:00
PG started right before: 2010-04-24 00:00:00
PG started right before: 2010-04-25 00:00:00
PG started right before: 2010-04-26 00:00:00
PG started right before: 2010-04-27 00:00:00
PG started right before: 2010-04-29 00:00:00
PG started right before: 2010-04-30 00:00:00
PG started right before: 2010-05-01 00:00:00
PG started right before: 2010-05-02 00:00:00
PG started right before: 2010-05-09 00:00:00
PG started right before: 2010-05-10 00:00:00
PG started right before: 2010-05-11 00:00:00
PG started right before: 2010-05-12 00:00:00
PG started right before: 2010-05-13 00:00:00
PG started right before: 2010-05-15 00:00:00
PG started right before: 2010-05-21 00:00:00
PG started right before: 2010-05-22 00:00:00
PG started right before: 2010-05-23 00:00:00
PG started right before: 2010-05-28 00:00:00
PG started right before: 2010-05-29 00:00:00
PG started right before: 2010-05-30 00:00:00
PG started right before: 2010-06-04 00:00:00
PG started right before: 2010-06-05 00:00:00
PG started right before: 2010-06-06 00:00:00
PG started right before: 2010-06-12 00:00:00
PG started right before: 2010-06-13 00:00:00
PG started right before: 2010-06-14 00:00:00
PG started right before: 2010-06-19 00:00:00
PG started right before: 2010-06-20 00:00:00
PG started right before: 2010-06-21 00:00:00
PG started right before: 2010-06-25 00:00:00
PG started right before: 2010-06-26 00:00:00
PG started right before: 2010-06-28 00:00:00
PG started right before: 2010-06-29 00:00:00
PG started right before: 2010-07-04 00:00:00
PG started right before: 2010-07-22 00:00:00
PG started right before: 2010-07-23 00:00:00
PG started right before: 2010-07-24 00:00:00
PG started right before: 2010-07-25 00:00:00
PG started right before: 2010-07-29 00:00:00
PG started right before: 2010-07-30 00:00:00
PG started right before: 2010-07-31 00:00:00
PG started right before: 2010-08-01 00:00:00
PG started right before: 2010-08-04 00:00:00
PG started right before: 2010-08-05 00:00:00
PG started right before: 2010-08-12 00:00:00
PG started right before: 2010-08-13 00:00:00
PG started right before: 2010-08-14 00:00:00
PG started right before: 2010-08-16 00:00:00
PG started right before: 2010-08-25 00:00:00
PG started right before: 2010-08-26 00:00:00
PG started right before: 2010-08-29 00:00:00
PG started right before: 2010-09-04 00:00:00
PG started right before: 2010-09-09 00:00:00
PG started right before: 2010-09-10 00:00:00
PG started right before: 2010-09-11 00:00:00
PG started right before: 2010-09-12 00:00:00
PG started right before: 2010-09-13 00:00:00
PG started right before: 2010-09-14 00:00:00
PG started right before: 2010-09-15 00:00:00
PG started right before: 2010-09-18 00:00:00
PG started right before: 2010-09-23 00:00:00
PG started right before: 2010-09-24 00:00:00
PG started right before: 2010-09-25 00:00:00
PG started right before: 2010-10-02 00:00:00
PG started right before: 2010-10-03 00:00:00
PG started right before: 2010-10-05 00:00:00
PG started right before: 2010-10-06 00:00:00
PG started right before: 2010-10-08 00:00:00
PG started right before: 2010-10-10 00:00:00
PG started right before: 2010-10-11 00:00:00
PG started right before: 2010-10-16 00:00:00
PG started right before: 2010-10-17 00:00:00
PG started right before: 2010-10-20 00:00:00
PG started right before: 2010-10-24 00:00:00
PG started right before: 2010-10-27 00:00:00
PG started right before: 2010-10-30 00:00:00
PG started right before: 2010-10-31 00:00:00
PG started right before: 2010-11-02 00:00:00

Mean: 147.61094715168153
Standard Deviation: 780.2908435877481
Variance: 608853.8005868795
1934/1934 [==============================] - 310s 158ms/step - loss: 0.0058
16/16 [==============================] - 3s 134ms/step
Test RMSE: 322.38036411034545
Test MAE: 86.72363813992908

PG started right before: 2009-07-03 00:00:00
PG started right before: 2009-07-04 00:00:00
PG started right before: 2009-07-05 00:00:00
PG started right before: 2009-07-08 00:00:00
PG started right before: 2009-07-09 00:00:00
PG started right before: 2009-07-13 00:00:00
PG started right before: 2009-07-14 00:00:00
PG started right before: 2009-07-15 00:00:00
PG started right before: 2009-07-16 00:00:00
PG started right before: 2009-07-17 00:00:00
PG started right before: 2009-07-18 00:00:00
PG started right before: 2009-07-19 00:00:00
PG started right before: 2009-07-20 00:00:00
PG started right before: 2009-07-21 00:00:00
PG started right before: 2009-07-22 00:00:00
PG started right before: 2009-07-25 00:00:00
PG started right before: 2009-08-25 00:00:00
PG started right before: 2009-08-26 00:00:00
PG started right before: 2009-08-27 00:00:00
PG started right before: 2009-08-28 00:00:00
PG started right before: 2009-08-29 00:00:00
PG started right before: 2009-08-30 00:00:00
PG started right before: 2009-08-31 00:00:00
PG started right before: 2009-09-07 00:00:00
PG started right before: 2009-09-08 00:00:00
PG started right before: 2009-09-09 00:00:00
PG started right before: 2009-09-20 00:00:00
PG started right before: 2009-09-21 00:00:00
PG started right before: 2009-09-22 00:00:00
PG started right before: 2009-09-23 00:00:00
PG started right before: 2009-09-24 00:00:00
PG started right before: 2009-09-25 00:00:00
PG started right before: 2009-09-28 00:00:00
PG started right before: 2009-10-05 00:00:00
PG started right before: 2009-10-06 00:00:00
PG started right before: 2009-10-07 00:00:00
PG started right before: 2009-10-08 00:00:00
PG started right before: 2010-04-11 00:00:00
PG started right before: 2010-04-12 00:00:00
PG started right before: 2010-04-13 00:00:00
PG started right before: 2010-04-14 00:00:00
PG started right before: 2010-04-15 00:00:00
PG started right before: 2010-04-16 00:00:00
PG started right before: 2010-04-17 00:00:00
PG started right before: 2010-04-18 00:00:00
PG started right before: 2010-04-25 00:00:00
PG started right before: 2010-05-01 00:00:00
PG started right before: 2010-05-02 00:00:00
PG started right before: 2010-05-03 00:00:00
PG started right before: 2010-05-07 00:00:00
PG started right before: 2010-05-08 00:00:00
PG started right before: 2010-05-09 00:00:00
PG started right before: 2010-05-10 00:00:00
PG started right before: 2010-05-14 00:00:00
PG started right before: 2010-05-16 00:00:00
PG started right before: 2010-05-17 00:00:00
PG started right before: 2010-05-18 00:00:00

Mean: 761.8822115991763
Standard Deviation: 2004.6482107288575
Variance: 4018614.44877841
1934/1934 [==============================] - 253s 129ms/step - loss: 0.0061
16/16 [==============================] - 3s 119ms/step
Test RMSE: 3639.499538276778
Test MAE: 1516.2712461583662

PG started right before: 2009-07-03 00:00:00
PG started right before: 2009-07-04 00:00:00
PG started right before: 2009-07-08 00:00:00
PG started right before: 2009-07-14 00:00:00
PG started right before: 2009-07-17 00:00:00
PG started right before: 2009-07-18 00:00:00
PG started right before: 2009-07-19 00:00:00
PG started right before: 2009-07-24 00:00:00
PG started right before: 2009-07-25 00:00:00
PG started right before: 2009-07-29 00:00:00
PG started right before: 2009-07-30 00:00:00
PG started right before: 2009-07-31 00:00:00
PG started right before: 2009-08-01 00:00:00
PG started right before: 2009-08-05 00:00:00
PG started right before: 2009-08-06 00:00:00
PG started right before: 2009-08-07 00:00:00
PG started right before: 2009-08-12 00:00:00
PG started right before: 2009-08-13 00:00:00
PG started right before: 2009-08-14 00:00:00
PG started right before: 2009-08-15 00:00:00
PG started right before: 2009-08-19 00:00:00
PG started right before: 2009-08-20 00:00:00
PG started right before: 2009-08-28 00:00:00
PG started right before: 2009-08-29 00:00:00
PG started right before: 2009-08-30 00:00:00
PG started right before: 2009-08-31 00:00:00
PG started right before: 2009-09-02 00:00:00
PG started right before: 2009-09-03 00:00:00
PG started right before: 2009-09-04 00:00:00
PG started right before: 2009-09-05 00:00:00
PG started right before: 2009-09-11 00:00:00
PG started right before: 2009-09-12 00:00:00
PG started right before: 2009-09-13 00:00:00
PG started right before: 2009-09-17 00:00:00
PG started right before: 2009-09-19 00:00:00
PG started right before: 2009-09-20 00:00:00
PG started right before: 2009-09-26 00:00:00
PG started right before: 2009-09-27 00:00:00
PG started right before: 2009-09-28 00:00:00
PG started right before: 2009-09-29 00:00:00
PG started right before: 2009-09-30 00:00:00
PG started right before: 2009-10-01 00:00:00
PG started right before: 2009-10-02 00:00:00
PG started right before: 2009-10-03 00:00:00
PG started right before: 2009-10-04 00:00:00
PG started right before: 2009-10-10 00:00:00
PG started right before: 2009-10-11 00:00:00
PG started right before: 2009-10-13 00:00:00
PG started right before: 2009-10-18 00:00:00
PG started right before: 2009-10-19 00:00:00
PG started right before: 2009-10-20 00:00:00
PG started right before: 2009-10-23 00:00:00
PG started right before: 2009-10-27 00:00:00
PG started right before: 2009-10-28 00:00:00
PG started right before: 2009-10-30 00:00:00
PG started right before: 2009-11-05 00:00:00
PG started right before: 2009-11-08 00:00:00
PG started right before: 2009-11-13 00:00:00
PG started right before: 2009-11-14 00:00:00
PG started right before: 2009-11-15 00:00:00
PG started right before: 2009-11-21 00:00:00
PG started right before: 2009-11-22 00:00:00
PG started right before: 2009-11-24 00:00:00
PG started right before: 2009-11-27 00:00:00
PG started right before: 2009-11-28 00:00:00
PG started right before: 2009-12-03 00:00:00
PG started right before: 2009-12-05 00:00:00
PG started right before: 2009-12-06 00:00:00
PG started right before: 2009-12-13 00:00:00
PG started right before: 2009-12-14 00:00:00
PG started right before: 2009-12-15 00:00:00
PG started right before: 2009-12-19 00:00:00
PG started right before: 2009-12-26 00:00:00
PG started right before: 2009-12-27 00:00:00
PG started right before: 2009-12-28 00:00:00
PG started right before: 2010-01-03 00:00:00
PG started right before: 2010-01-04 00:00:00
PG started right before: 2010-01-15 00:00:00
PG started right before: 2010-01-16 00:00:00
PG started right before: 2010-01-17 00:00:00
PG started right before: 2010-01-18 00:00:00
PG started right before: 2010-01-23 00:00:00
PG started right before: 2010-01-24 00:00:00
PG started right before: 2010-01-25 00:00:00
PG started right before: 2010-01-26 00:00:00
PG started right before: 2010-01-27 00:00:00
PG started right before: 2010-01-29 00:00:00
PG started right before: 2010-01-30 00:00:00
PG started right before: 2010-01-31 00:00:00
PG started right before: 2010-02-01 00:00:00
PG started right before: 2010-02-02 00:00:00
PG started right before: 2010-02-05 00:00:00
PG started right before: 2010-02-09 00:00:00
PG started right before: 2010-02-10 00:00:00
PG started right before: 2010-02-11 00:00:00
PG started right before: 2010-02-12 00:00:00
PG started right before: 2010-02-15 00:00:00
PG started right before: 2010-02-23 00:00:00
PG started right before: 2010-02-24 00:00:00
PG started right before: 2010-02-25 00:00:00
PG started right before: 2010-02-26 00:00:00
PG started right before: 2010-02-27 00:00:00
PG started right before: 2010-03-02 00:00:00
PG started right before: 2010-03-07 00:00:00
PG started right before: 2010-03-08 00:00:00
PG started right before: 2010-03-10 00:00:00
PG started right before: 2010-03-11 00:00:00
PG started right before: 2010-03-12 00:00:00
PG started right before: 2010-03-16 00:00:00
PG started right before: 2010-03-17 00:00:00
PG started right before: 2010-03-18 00:00:00
PG started right before: 2010-03-19 00:00:00
PG started right before: 2010-05-03 00:00:00
PG started right before: 2010-05-04 00:00:00
PG started right before: 2010-05-22 00:00:00
PG started right before: 2010-05-23 00:00:00
PG started right before: 2010-05-29 00:00:00
PG started right before: 2010-05-30 00:00:00
PG started right before: 2010-05-31 00:00:00
PG started right before: 2010-06-10 00:00:00
PG started right before: 2010-06-11 00:00:00
PG started right before: 2010-06-26 00:00:00
PG started right before: 2010-06-27 00:00:00
PG started right before: 2010-06-28 00:00:00
PG started right before: 2010-07-22 00:00:00
PG started right before: 2010-07-23 00:00:00
PG started right before: 2010-07-24 00:00:00
PG started right before: 2010-07-27 00:00:00
PG started right before: 2010-07-28 00:00:00
PG started right before: 2010-07-29 00:00:00
PG started right before: 2010-08-15 00:00:00
PG started right before: 2010-08-16 00:00:00
PG started right before: 2010-08-17 00:00:00
PG started right before: 2010-08-19 00:00:00
PG started right before: 2010-08-20 00:00:00
PG started right before: 2010-08-21 00:00:00
PG started right before: 2010-08-27 00:00:00
PG started right before: 2010-08-30 00:00:00
PG started right before: 2010-08-31 00:00:00
PG started right before: 2010-09-01 00:00:00
PG started right before: 2010-09-24 00:00:00
PG started right before: 2010-09-25 00:00:00

Name		Name	Last commit message	Last commit date
Latest commit History 27 Commits
.ipynb_checkpoints		.ipynb_checkpoints
README_imgres		README_imgres
datasets		datasets
DatasetAnalysis.ipynb		DatasetAnalysis.ipynb
README.md		README.md

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Repository files navigation

Using Time Series Predictive Models for Early Detection of Gambling Addiction in Problem Gamblers

Data Gathering

Data Transformation and Cleaning

Merging of Datasets

Feature Selection

Model Fitting

K - Means Clustering

K means Evaluation

K means Labeling Analysis

Filtering of data based on hypothesis testing for Stationarity

Model Fitting

ARIMA/SARIMA model fitting

LSTM for non stationary users

GRID SEARCH FOR BEST PARAMS

Saved results of LSTM Hyperparameter tuning for better reducing computation time.

Model Fitting of LSTM and Evaluation in Loop

About

Releases

Packages

Languages

ramxbx/GamblingAddictionTimeSeriesPrediction

Folders and files

Latest commit

History

Repository files navigation

Using Time Series Predictive Models for Early Detection of Gambling Addiction in Problem Gamblers

Data Gathering

Data Transformation and Cleaning

Merging of Datasets

Feature Selection

Model Fitting

K - Means Clustering

K means Evaluation

K means Labeling Analysis

Filtering of data based on hypothesis testing for Stationarity

Model Fitting

ARIMA/SARIMA model fitting

LSTM for non stationary users

GRID SEARCH FOR BEST PARAMS

Saved results of LSTM Hyperparameter tuning for better reducing computation time.

Model Fitting of LSTM and Evaluation in Loop

About

Topics

Resources

Stars

Watchers

Forks

Releases

Packages 0

Languages

Packages