-
Notifications
You must be signed in to change notification settings - Fork 0
/
jobutil.py
223 lines (205 loc) · 8.14 KB
/
jobutil.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
##
#
##
import sys
import os
import datetime as dt
import numpy as np
import pandas as pd
from sklearn.model_selection import train_test_split
from imblearn.under_sampling import RandomUnderSampler
import database as db
import util
import models as M
np.random.seed(7)
os.system("module2 load matlab")
def build_all_classifiers(goes):
# Dataset
if goes == "1": _xparams,X,y = db.load_data_for_deterministic_bin_clf()
else: _xparams,X,y = db.load_data_with_goes_for_deterministic_bin_clf()
rus = RandomUnderSampler(return_indices=True)
X_resampled, y_resampled, idx_resampled = rus.fit_sample(X, y)
#X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=1.0/3.0, random_state=42)
X_train, X_test, y_train, y_test = train_test_split(X_resampled, y_resampled, test_size=1.0/3.0, random_state=42)
# Initialize metrix
CLF = util.get_classifiers()
roc_eval_details = {}
feature_importance = {}
_C2x2 = {}
for I,clfs in enumerate(CLF):
gname = clfs["name"]
roc_eval_details[gname] = {}
for clf,name in clfs["methods"]:
print("--> Running '%s' model" % name)
clf.fit(X_train, y_train)
if hasattr(clf,"feature_importances_"):
feature_importance[name] = dict(zip(_xparams,clf.feature_importances_))
pass
if hasattr(clf, "predict_proba") or hasattr(clf, "decision_function"):
roc_eval_details[gname][name] = {}
roc_eval_details[gname][name]["y_score"], roc_eval_details[gname][name]["fpr"], roc_eval_details[gname][name]["tpr"], roc_eval_details[gname][name]["roc_auc"] = util.get_roc_details(clf, X_test, y_test)
roc_eval_details[gname][name]["c"] = np.random.rand(3,1).ravel().tolist()
pass
_C2x2[name] = util.validate_model_matrices(clf, X_test, y_test)
pass
pass
util.plot_deterministic_roc_curves(roc_eval_details, goes)
for name in _C2x2.keys():
print("Running '%s' model" % name)
print("==================")
_C2x2[name].summary(verbose=True)
print("**************************************************")
print("\n\n")
pass
return
def build_lstm_classification(goes):
if goes == 0: isgoes = False
else: isgoes = True
rus = RandomUnderSampler(return_indices=True)
_xparams,X,y = db.load_data_with_goes_for_lstm_bin_clf(isgoes = False)
X_resampled, y_resampled, idx_resampled = rus.fit_sample(X, y)
X_train, X_test, y_train, y_test = train_test_split(X_resampled, y_resampled, test_size=1.0/3.0, random_state=42)
clf = util.get_lstm_classifier(X.shape[1])
print clf
# clf.fit(X_train, y_train, batch_size=64, nb_epoch=10, validation_split = 0.1, verbose = 1)
return
#def build_all_regressor():
# _o, _xparams, _yparam = db.load_data_for_deterministic_reg()
# _oh = _o[_o[_yparam[0]] >= 5.5]
# _ol = _o[_o[_yparam[0]] <= 5.5]
# Xh = _oh.as_matrix(_xparams)
# Xl = _ol.as_matrix(_xparams)
# yh = np.array(_oh[_yparam[0]].tolist()).reshape((len(_oh),1))
# yl = np.array(_ol[_yparam[0]].tolist()).reshape((len(_ol),1))
# f_clf = "out/rf.pkl"
# clf = util.get_best_determinsistic_classifier(f_clf)
# print("--> Classifier loaded..")
# print(Xh.shape,yh.shape,Xl.shape,yl.shape)
# Xh_train, Xh_test, yh_train, yh_test = train_test_split(Xh, yh, test_size=1.0/3.0, random_state=42)
# Xl_train, Xl_test, yl_train, yl_test = train_test_split(Xl, yl, test_size=1.0/3.0, random_state=42)
# prh = clf.predict_proba(Xh_test)[:,0]
# prl = clf.predict_proba(Xl_test)[:,0]
# y_obs = []
# y_obs.extend(yh_test[:,0].tolist())
# y_obs.extend(yl_test[:,0].tolist())
# for r in regs:
# yl_pred = []
# yh_pred = []
# y_pred = []
# print("Model:"+r)
# regO = util.get_regressor(r, 0)
# regL = util.get_regressor(r, 0)
# RL = regL[0]
# RL.fit(Xl_train,yl_train)
# RO = regO[0]
# RO.fit(Xh_train,yh_train)
# model = r
# for I,p in enumerate(prh):
# if p < .7: yh_pred.append(RL.predict(Xh_test[I,:].reshape((1,10))).ravel()[0])
# else: yh_pred.append(RO.predict(Xh_test[I,:].reshape((1,10))).ravel()[0])
# pass
# for I,p in enumerate(prl):
# if p < .7: yl_pred.append(RL.predict(Xl_test[I,:].reshape((1,10))).ravel()[0])
# else: yl_pred.append(RO.predict(Xl_test[I,:].reshape((1,10))).ravel()[0])
# pass
# y_pred.extend(yh_pred)
# y_pred.extend(yl_pred)
# _eval_details = util.run_validation(y_pred,y_obs,"[1995-2016]",model)
# print _eval_details
# #break
# pass
# return
def run_deterministic_clf_reg_model(args):
if len(args) == 0: print "python jobutil.py 2 <reg model> <trw> <year>(1995-2016)"
else:
model = args[0]
trw = int(args[1])
if model == "dtree" or model == "etree" or model == "knn" or model == "ada":
if args[2] == "all":
for y in range(1995,2017): M.run_model_process_based_on_deterministic_algoritms(y, model, trw=trw)
pass
else:
y = int(args[2])
M.run_model_process_based_on_deterministic_algoritms(y, model, trw=trw)
pass
else:
if args[2] == "all":
for y in range(1995,2017): M.run_model_process_based_on_deterministic_algoritms(y, model, trw=trw)#M.run_model_based_on_deterministic_algoritms(y, model, trw=trw)
pass
else:
y = int(args[2])
#M.run_model_based_on_deterministic_algoritms(y, model, trw=trw)
M.run_model_process_based_on_deterministic_algoritms(y, model, trw=trw)
pass
pass
return
def run_gp_clf_reg_model(args):
if len(args) == 0: print "python jobutil.py 3 GPR <ktype>(RBF/RQ/Matern) <trw> <year>(1995-2016)"
else:
model = args[0]
ktype = args[1]
trw = int(args[2])
if args[3] == "all":
for y in range(1995,2017): M.run_model_based_on_gp(y, kt=ktype, trw = trw)
pass
else:
y = int(args[3])
M.run_model_based_on_gp(y, kt=ktype, trw = trw)
pass
return
def run_lstm_clf_reg_model(args):
if len(args) == 0: print "python jobutil.py 5 LSTM <trw> <year>(1995-2016)"
else:
model = args[0]
trw = int(args[1])
if args[2] == "all":
for y in range(2001,2017): M.run_model_based_on_lstm(y, model="LSTM", trw = trw)
pass
else:
y = int(args[2])
M.run_model_based_on_lstm(y, model="LSTM", trw = trw)
pass
return
def run_lstmgp_clf_reg_model(args):
if len(args) == 0: print "python jobutil.py 7 deepGP <trw> <year>(1995-2016)"
else:
model = args[0]
trw = int(args[1])
i = int(args[3])
if args[2] == "all":
for y in range(1995,2017): M.run_model_based_on_deepgp(y, model="deepGP", trw = trw, i = 0)
pass
else:
y = int(args[2])
M.run_model_based_on_deepgp(y, model="deepGP", trw = trw,i=i)
pass
return
def run_model_stats(args):
if len(args) == 0: print "python jobutil.py 4 <reg model> <trw>"
else:
model = args[0]
trw = int(args[1])
util.get_stats(model, trw)
return
def run_tss_plot(args):
if len(args) == 0: print "python jobutil.py 6 <reg model> <trw>"
else:
model = args[0]
trw = int(args[1])
util.run_for_TSS(model, trw)
return
if __name__ == "__main__":
args = sys.argv[1:]
if len(args) == 0: print "Invalid call sequence!! python jobutil.py {1/2/3...}"
else:
ctx = int(args[0])
if ctx == 1: build_all_classifiers(args[1])
if ctx == 2: run_deterministic_clf_reg_model(args[1:])
if ctx == 3: run_gp_clf_reg_model(args[1:])
if ctx == 4: run_model_stats(args[1:])
if ctx == 5: run_lstm_clf_reg_model(args[1:])
if ctx == 7: run_lstmgp_clf_reg_model(args[1:])
if ctx == 9: build_lstm_classification(args[1])
if ctx == 6: run_tss_plot(args[1:])
pass
pass