-
Notifications
You must be signed in to change notification settings - Fork 1
/
utils.py
251 lines (206 loc) · 6.77 KB
/
utils.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
def tr_vl_split(X, Y, ratio=0.25):
""" Splits a dataset into two parts with random elements
Parameters
----------
X
The features to be splitted
Y
The targets to be splitted
ratio : float, optional
The ratio of the split
"""
import numpy as np
import math
ix = np.random.randint(low = 0, high = len(X), size = math.floor(ratio * len(X)))
X_vl, Y_vl = X[ix], Y[ix]
X_tr, Y_tr = np.delete(X, ix, axis = 0), np.delete(Y, ix, axis = 0)
return X_tr, X_vl, Y_tr, Y_vl
def plot_and_save(title, history, validation_history=None, ylabel="Loss", xlabel="Epochs", savefile=None):
""" Plots some data and saves the image
Parameters
----------
title : str
Title to be printed on top of the plot
history : list
The values to be printed
validation_history : list, optional
The values of the validation loss to be printed alongside the history
ylabel : str, optional
The label of the y axis
xlabel : str, optional
The label of the x axis
savefile : str, optional
The name of the file where to save the plot, in the plot folder
"""
import matplotlib.pyplot as plot
fig, ax = plot.subplots()
ax.plot(history, label=ylabel)
if not(validation_history is None): ax.plot(validation_history, label='Validation Loss')
ax.set_ylabel(ylabel)
ax.set_xlabel(xlabel)
ax.set_title(title)
ax.legend()
plot.gca().margins(x=0)
fig.set_size_inches(18.5, 10.5)
if not(savefile is None): plot.savefig("plots/" + savefile + ".png")
plot.clf()
def multiline_plot(title, histories, legend_names, ylabel="Loss", xlabel="Epochs", style="dark", showlegend=True, showgrid=False, savefile=None, alternateDots=False, prefix=""):
""" Plots multiple data curves on the same cartesian plane
Parameters
----------
title : str
Title to be printed on top of the plot
histories : list
The list of cuves to be printed
legend_names : list
The list of legend names to be printed. One for each history.
ylabel : str, optional
The label of the y axis
xlabel : str, optional
The label of the x axis
style: string, optional
Name of the seaborn style to be applied
showlegend: boolean
Show or hide legendnames
showgrid: boolean
Show or hide gridlines on the plot background. ]
savefile : str, optional
The name of the file where to save the plot, in the plot folder
"""
import matplotlib.pyplot as plt
import seaborn as sns
l = len(histories)
plt.rcParams.update({'font.size': 18})
sns.set()
with sns.color_palette(style, n_colors=l):
fig, ax = plt.subplots()
for i in reversed(range(l)):
lStyle = '-' if ( alternateDots and i % 2 == 0) else ':'
ax.plot(histories[i], label=legend_names[i], linestyle=lStyle, linewidth=2)
ax.set_ylabel(ylabel)
ax.set_xlabel(xlabel)
ax.set_title(title)
if (showlegend): ax.legend()
if (showgrid): ax.grid(linestyle='--')
plt.gca().margins(x=0)
fig.set_size_inches(12, 8)
if not(savefile is None): plt.savefig(prefix + "plots/" + savefile + ".png")
plt.clf()
log(prefix + "logs/" + savefile, histories)
return
def roc_curve(title, FPR, TPR, AUC, xlabel="Specificity", ylabel="Sensitivity", savefile=None):
""" Plots the ROC curve
Parameters
----------
title : str
Title to be printed on top of the plot
FPR : list
The false positive rate, x-axis
TPR : list
The true positive rate, y-axis
AUC : float
The area under the ROC curve
xlabel : str, optional
The label of the x axis
ylabel : str, optional
The label of the y axis
savefile : str, optional
The name of the file where to save the plot, in the plot folder
"""
import matplotlib.pyplot as plot
fig, ax = plot.subplots()
ax.plot(FPR, TPR)
ax.plot([0, 1], [0, 1],'r--')
ax.plot([], [], ' ', label="AUC = " + str(AUC))
ax.set_xlabel(xlabel)
ax.set_ylabel(ylabel)
ax.set_title(title)
plot.gca().margins(x=0)
plot.xlim([-0.01,1.01])
plot.ylim([-0.01,1.01])
plot.legend()
fig.set_size_inches(18.5, 10.5)
if not(savefile is None): plot.savefig("plots/" + savefile + ".png")
plot.clf()
# Only for MONK
def confusion_matrix(title="sample", values = (0,0,0,0), savefile=None):
""" Plots the confusion matrix
Parameters
----------
title : str
Title to be printed on top of the plot
TP : int
The true positives
FN : int
The false negatives
FP : int
The false positives
TN : int
The true negatives
xlabel : str, optional
The label of the x axis
ylabel : str, optional
The label of the y axis
savefile : str, optional
The name of the file where to save the plot, in the plot folder
"""
import seaborn as sn
import pandas as pd
import matplotlib.pyplot as plot
TP, TN, FP, FN = values
cfm = [[TP,FP],
[FN,TN]]
x_labels = ["Predicted 1", "Predicted 2"]
y_labels = ["Actually 1", "Actually 2" ]
df_cm = pd.DataFrame(cfm, range(2), range(2))
sn.set(font_scale=1.4) # for label size
sn.heatmap(df_cm, annot=True, fmt="g", cmap="Blues" ,annot_kws={"size": 16}, xticklabels=x_labels, yticklabels=y_labels).set(title=title)
plot.gca().margins(x=0)
if not(savefile is None): plot.savefig("plots/" + savefile + ".png")
plot.clf()
def log(filename, data):
""" Saves some data in a pickle file in the logs folder
Parameters
----------
filename : str
The name of the file where to save
data
The data to be saved
"""
import pickle
with open(filename + ".pkl", "wb") as logfile:
pickle.dump(data, logfile)
def compare(a, b, tollerance=1e-03):
""" Compares two numbers with some tollerance
Parameters
----------
a : float
The first number to be compared
b : float
The second number to be compared
tollerance : float, optional
The tollerance of the comparison
Returns
-------
bool
True if a and b are equal up to the tollerance
"""
return abs(a - b) <= tollerance * max(abs(a), abs(b))
def shuffle(a,b):
""" Shuffles two lists in parallel
Parameters
----------
a : list
The first list to be shuffled
b : list
The second list to be shuffled
Returns
-------
list, list
The two lists, shuffled
"""
import numpy as np
assert len(a) == len(b)
randomize = np.arange(len(a))
np.random.shuffle(randomize)
return a[randomize], b[randomize]