-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathOneShot_NewAnalysis.py
731 lines (571 loc) · 35 KB
/
OneShot_NewAnalysis.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
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
# -*- coding: utf-8 -*-
"""
Created on Fri Aug 17 11:16:03 2018
@author: Adam
"""
import numbers
import numpy as np
import pandas as pd
import sklearn
from scipy.stats import kurtosis
from scipy.stats import skew
from sklearn.ensemble.forest import RandomForestClassifier
from sklearn.tree import DecisionTreeClassifier
from sklearn.neural_network import MLPClassifier
from keras.layers import Input, Dense
from keras.models import Model
from datetime import datetime
# Model and feature selection
from sklearn.feature_selection import SelectKBest
from sklearn.model_selection import KFold
from sklearn.feature_selection import chi2
# Classification metrics
from sklearn.metrics import f1_score
from sklearn.metrics import precision_score
from sklearn.metrics import recall_score
from sklearn.metrics import accuracy_score
from sklearn import preprocessing
from PersonalClassifier import PersonalClassifier
from sklearn.ensemble import AdaBoostClassifier
from sklearn.ensemble import VotingClassifier
from sklearn.svm import SVC
from sklearn.linear_model import logistic
def _convert_prediction(X):
X.loc[X['Prediction']==1,"VotePrediction"] = X.loc[X['Prediction']==1,"Pref1"]
X.loc[X['Prediction']==2,"VotePrediction"] = X.loc[X['Prediction']==2,"Pref2"]
X.loc[X['Prediction']==3,"VotePrediction"] = X.loc[X['Prediction']==3,"Pref3"]
return X
def _generate_action_name(X):
# Generate action name
# Action mapping table
d = {'scenario': ['A', 'A', 'A', 'B', 'B', 'B', 'C', 'C', 'C', 'D', 'D', 'D', 'E', 'E', 'E', 'F', 'F', 'F'],
'action': [1, 2, 3, 1, 2, 3, 1, 2, 3, 1, 2, 3, 1, 2, 3, 1, 2, 3],
'action_name': ['TRT', 'DOM', 'DOM', 'TRT', 'DOM', 'DOM', 'TRT', 'WLB', 'DOM', 'TRT', 'DOM', 'SLB', 'TRT',
'CMP\WLB', 'DOM', 'TRT', 'CMP', 'SLB']}
action_map_df = pd.DataFrame(data=d)
X['Action_name'] = [(action_map_df.loc[(action_map_df.scenario == str(x[1]['Scenario'])) & (
action_map_df.action == int(X.loc[x[0], 'Action'])), 'action_name']).values[0] for x in
X.iterrows()]
return X
def _data_cleaning(data_df,is_target ,le):
if is_target:
data_df = data_df.astype("category")
data_df = le.fit_transform(data_df)
else:
for c in data_df.columns:
if data_df[c].dtype in (object, str, np.object, bool):
if not (data_df[c].dtype in (int, float)):
data_df[c] = le.fit_transform(data_df[c])
return data_df
def _generate_A_ratios(X, X_train, y_train, voter):
"""Generate A ratios - That is TRT-ratio, CMP-ratio, WLB-ratio, SLB-ratio, DOM-ratio
Action is in {TRT,DLB,SLB,WLB,CMP,DOM}
Scenario is in {A,B,C,D,E,F}
"""
start_ind_col = len(X.columns)
availability_counter = np.count_nonzero([x[1].Scenario in ['A','B','C','D','E','F'] for x in X_train.loc[X_train['VoterID'] == voter.VoterID].iterrows()])
X.loc[X['VoterID'] == voter.VoterID, 'TRT-ratio'] = np.count_nonzero((['TRT' in x[1]['Action_name'] for x in X_train.loc[X_train['VoterID'] == voter.VoterID].iterrows()]))/availability_counter if availability_counter>0 else 0
availability_counter = np.count_nonzero([x[1].Scenario in ['C','E'] for x in X_train.loc[X_train['VoterID'] == voter.VoterID].iterrows()])
X.loc[X['VoterID'] == voter.VoterID, 'WLB-ratio'] = np.count_nonzero((['WLB' in x[1]['Action_name'] for x in X_train.loc[X_train['VoterID'] == voter.VoterID].iterrows()]))/availability_counter if availability_counter>0 else 0
availability_counter = np.count_nonzero([x[1].Scenario in ['D','F'] for x in X_train.loc[X_train['VoterID'] == voter.VoterID].iterrows()])
X.loc[X['VoterID'] == voter.VoterID, 'SLB-ratio'] = np.count_nonzero((['SLB' in x[1]['Action_name'] for x in X_train.loc[X_train['VoterID'] == voter.VoterID].iterrows()]))/availability_counter if availability_counter>0 else 0
#X.loc[X['VoterID'] == voter.VoterID, 'LB-ratio'] = np.count_nonzero((['LB' in x[1]['Action_name'] for x in X_train.loc[X_train['VoterID'] == voter.VoterID].iterrows()]))/np.count_nonzero([x[1].Scenario in ['C','D','E','F'] for x in X_train.loc[X_train['VoterID'] == voter.VoterID].iterrows()])
availability_counter = np.count_nonzero([x[1].Scenario in ['E','F'] for x in X_train.loc[X_train['VoterID'] == voter.VoterID].iterrows()])
X.loc[X['VoterID'] == voter.VoterID, 'CMP-ratio'] = np.count_nonzero((['CMP' in x[1]['Action_name'] for x in X_train.loc[X_train['VoterID'] == voter.VoterID].iterrows()]))/availability_counter if availability_counter>0 else 0
availability_counter = np.count_nonzero([x[1].Scenario in ['A','B','C','D','E','F'] for x in X_train.loc[X_train['VoterID'] == voter.VoterID].iterrows()])
X.loc[X['VoterID'] == voter.VoterID, 'DOM-ratio'] = np.count_nonzero((['DOM' in x[1]['Action_name'] for x in X_train.loc[X_train['VoterID'] == voter.VoterID].iterrows()]))/availability_counter if availability_counter>0 else 0
X.loc[X['VoterID'] == voter.VoterID, 'DOM-counter'] = np.count_nonzero((['DOM' in x[1]['Action_name'] for x in X_train.loc[X_train['VoterID'] == voter.VoterID].iterrows()]))
X['TRT-ratio'] = (X['TRT-ratio']).astype(float)
X['WLB-ratio'] = (X['WLB-ratio']).astype(float)
X['SLB-ratio'] = (X['SLB-ratio']).astype(float)
X['CMP-ratio'] = (X['CMP-ratio']).astype(float)
X['DOM-ratio'] = (X['DOM-ratio']).astype(float)
X['DOM-counter'] = (X['DOM-counter']).astype(float)
end_ind_col = len(X.columns)
return X, list(range(start_ind_col, end_ind_col))
def _generate_is_random_voter(X):
"""Identify random voters using the rule of DOM-counter >= 2 (excluding SLB actions)"""
X['Is_Random'] = [x >= 2 for x in X['DOM-counter']]
return X
def _generate_voter_type(X):
"""Generate Voter Type using thresholds over the A-ratio values"""
X['VoterType'] = 'Other'
#X.loc[ [int(x[1]['CMP-ratio'])>=0.7 for x in X.iterrows()], 'VoterType'] = 'CMP'
X.loc[ [float(x[1]['WLB-ratio'])>0.8 for x in X.iterrows()], 'VoterType'] = 'LB'
X.loc[ [float(x[1]['TRT-ratio'])>0.9 for x in X.iterrows()], 'VoterType'] = 'TRT'
return X
def _generate_pref_gaps(X):
X["VotesPref1PreVote"] = [x[1]["VotesCand"+str(x[1]["Pref1"])+"PreVote"] for x in X.iterrows()]
X["VotesPref2PreVote"] = [x[1]["VotesCand"+str(x[1]["Pref2"])+"PreVote"] for x in X.iterrows()]
X["VotesPref3PreVote"] = [x[1]["VotesCand"+str(x[1]["Pref3"])+"PreVote"] for x in X.iterrows()]
return X
def _generate_gaps(X):
"""Generate Gaps features"""
X['VotesLeader_poll'] = X[['VotesCand1PreVote','VotesCand2PreVote','VotesCand3PreVote']].max(axis = 1)
X['VotesRunnerup_poll'] = X[['VotesCand1PreVote','VotesCand2PreVote','VotesCand3PreVote']].apply(np.median, axis=1)
X['VotesLoser_poll'] = X[['VotesCand1PreVote','VotesCand2PreVote','VotesCand3PreVote']].min(axis = 1)
X['GAP12_poll'] = X['VotesLeader_poll'] - X['VotesRunnerup_poll']
X['GAP23_poll'] = X['VotesRunnerup_poll'] - X['VotesLoser_poll']
X['GAP13_poll'] = X['VotesLeader_poll'] - X['VotesLoser_poll']
#Preference based gaps - I think more suitable for ML for it's more synchronized across the scenarios
# for x in X.iterrows():
# vote = x[1]
# candPref1 = 'VotesCand' + str(vote.Pref1) + 'PreVote'
# candPref2 = 'VotesCand' + str(vote.Pref2) + 'PreVote'
# candPref3 = 'VotesCand' + str(vote.Pref3) + 'PreVote'
# X.loc[x[0], 'VotesPref1_poll'] = X.loc[x[0], candPref1]
# X.loc[x[0], 'VotesPref2_poll'] = X.loc[x[0], candPref2]
# X.loc[x[0], 'VotesPref3_poll'] = X.loc[x[0], candPref3]
#
#
X['GAP12_pref_poll'] = X['VotesPref1PreVote'] - X['VotesPref2PreVote']
X['GAP23_pref_poll'] = X['VotesPref2PreVote'] - X['VotesPref3PreVote']
X['GAP13_pref_poll'] = X['VotesPref1PreVote'] - X['VotesPref3PreVote']
return X
def _generate_scenario_type(X):
#initialize
X['TRT'] = 0
X['WLB'] = 0
X['SLB'] = 0
X['CMP'] = 0
X.loc[[x[1].Scenario in ['A','B','C','D','E','F'] for x in X],'TRT'] = 1
X.loc[[x[1].Scenario in ['C','E'] for x in X],'WLB'] = 1
X.loc[[x[1].Scenario in ['D','F'] for x in X],'SLB'] = 1
X.loc[[x[1].Scenario in ['E','F'] for x in X],'CMP'] = 1
return X
def _generate_feature_aggregation_class_dependant(X, X_train, y_train, voter, feature_name, aggregation_func):
action1_list = [ float(x[1][feature_name]) for x in X_train.loc[(X_train['VoterID'] == voter.VoterID) & (y_train == 1)].iterrows()]
if len(action1_list)>0:
X.loc[X['VoterID'] == voter.VoterID, feature_name + '_action1_' + aggregation_func.__name__] = aggregation_func(action1_list)
action2_list = [ float(x[1][feature_name]) for x in X_train.loc[(X_train['VoterID'] == voter.VoterID) & (y_train == 2)].iterrows()]
if len(action2_list)>0:
X.loc[X['VoterID'] == voter.VoterID, feature_name + '_action2_' + aggregation_func.__name__] = aggregation_func(action2_list)
action3_list = [ float(x[1][feature_name]) for x in X_train.loc[(X_train['VoterID'] == voter.VoterID) & (y_train == 3)].iterrows()]
if len(action3_list)>0:
X.loc[X['VoterID'] == voter.VoterID, feature_name + '_action3_' + aggregation_func.__name__] = aggregation_func(action3_list)
return X
def _generate_action_aggregation_features(X, X_train, y_train, voter):
aggregators = [np.average, np.std, np.median]
feature_name = "Action"
for aggregation_func in aggregators:
X.loc[X['VoterID'] == voter.VoterID, feature_name + "_" + aggregation_func.__name__] = aggregation_func([float(y_train[x[0]]) for x in X_train.loc[(X_train['VoterID'] == voter.VoterID) & (X_train['Scenario'].isin(['C','D','E','F']))].iterrows()])
return X
def _generate_gaps_features(X, X_train, y_train, voter):
start_ind_col = len(X.columns)
feature12 = 'GAP12_pref_poll'
feature23 = 'GAP23_pref_poll'
feature13 = 'GAP13_pref_poll'
features = [feature12, feature23, feature13]
aggregators = [np.average, np.std, np.median, np.min, np.max, skew, kurtosis]
scenarios = ['C','D','E','F']
for aggregator in aggregators:
for feature in features:
X = _generate_feature_aggregation_class_dependant(X, X_train.loc[X_train['Scenario'].isin(scenarios)], y_train, voter, feature, aggregator)
end_ind_col = len(X.columns)
return X, list(range(start_ind_col, end_ind_col))
def _generate_gap_dif_features(X):
start_ind_col = len(X.columns)
feature12 = 'GAP12_pref_poll'
feature23 = 'GAP23_pref_poll'
feature13 = 'GAP13_pref_poll'
X[feature12 + '_action1_' + np.average.__name__ + '_dif'] = X[feature12] - X[feature12 + '_action1_' + np.average.__name__ ]
X[feature12 + '_action2_' + np.average.__name__ + '_dif'] = X[feature12] - X[feature12 + '_action2_' + np.average.__name__ ]
X[feature12 + '_action3_' + np.average.__name__ + '_dif'] = X[feature12] - X[feature12 + '_action3_' + np.average.__name__ ]
X[feature23 + '_action1_' + np.average.__name__ + '_dif'] = X[feature23] - X[feature23 + '_action1_' + np.average.__name__ ]
X[feature23 + '_action2_' + np.average.__name__ + '_dif'] = X[feature23] - X[feature23 + '_action2_' + np.average.__name__ ]
X[feature23 + '_action3_' + np.average.__name__ + '_dif'] = X[feature23] - X[feature23 + '_action3_' + np.average.__name__ ]
X[feature13 + '_action1_' + np.average.__name__ + '_dif'] = X[feature13] - X[feature13 + '_action1_' + np.average.__name__ ]
X[feature13 + '_action2_' + np.average.__name__ + '_dif'] = X[feature13] - X[feature13 + '_action2_' + np.average.__name__ ]
X[feature13 + '_action3_' + np.average.__name__ + '_dif'] = X[feature13] - X[feature13 + '_action3_' + np.average.__name__ ]
X[feature12 + '_action1_' + np.median.__name__ + '_dif'] = X[feature12] - X[feature12 + '_action1_' + np.median.__name__ ]
X[feature12 + '_action2_' + np.median.__name__ + '_dif'] = X[feature12] - X[feature12 + '_action2_' + np.median.__name__ ]
X[feature12 + '_action3_' + np.median.__name__ + '_dif'] = X[feature12] - X[feature12 + '_action3_' + np.median.__name__ ]
X[feature23 + '_action1_' + np.median.__name__ + '_dif'] = X[feature23] - X[feature23 + '_action1_' + np.median.__name__ ]
X[feature23 + '_action2_' + np.median.__name__ + '_dif'] = X[feature23] - X[feature23 + '_action2_' + np.median.__name__ ]
X[feature23 + '_action3_' + np.median.__name__ + '_dif'] = X[feature23] - X[feature23 + '_action3_' + np.median.__name__ ]
X[feature13 + '_action1_' + np.median.__name__ + '_dif'] = X[feature13] - X[feature13 + '_action1_' + np.median.__name__ ]
X[feature13 + '_action2_' + np.median.__name__ + '_dif'] = X[feature13] - X[feature13 + '_action2_' + np.median.__name__ ]
X[feature13 + '_action3_' + np.median.__name__ + '_dif'] = X[feature13] - X[feature13 + '_action3_' + np.median.__name__ ]
X[feature12 + '_action1_' + np.max.__name__ + '_dif'] = X[feature12] - X[feature12 + '_action1_' + np.max.__name__ ]
X[feature12 + '_action2_' + np.max.__name__ + '_dif'] = X[feature12] - X[feature12 + '_action2_' + np.max.__name__ ]
X[feature12 + '_action3_' + np.max.__name__ + '_dif'] = X[feature12] - X[feature12 + '_action3_' + np.max.__name__ ]
X[feature23 + '_action1_' + np.max.__name__ + '_dif'] = X[feature23] - X[feature23 + '_action1_' + np.max.__name__ ]
X[feature23 + '_action2_' + np.max.__name__ + '_dif'] = X[feature23] - X[feature23 + '_action2_' + np.max.__name__ ]
X[feature23 + '_action3_' + np.max.__name__ + '_dif'] = X[feature23] - X[feature23 + '_action3_' + np.max.__name__ ]
X[feature13 + '_action1_' + np.max.__name__ + '_dif'] = X[feature13] - X[feature13 + '_action1_' + np.max.__name__ ]
X[feature13 + '_action2_' + np.max.__name__ + '_dif'] = X[feature13] - X[feature13 + '_action2_' + np.max.__name__ ]
X[feature13 + '_action3_' + np.max.__name__ + '_dif'] = X[feature13] - X[feature13 + '_action3_' + np.max.__name__ ]
X[feature12 + '_action1_' + np.min.__name__ + '_dif'] = X[feature12] - X[feature12 + '_action1_' + np.min.__name__ ]
X[feature12 + '_action2_' + np.min.__name__ + '_dif'] = X[feature12] - X[feature12 + '_action2_' + np.min.__name__ ]
X[feature12 + '_action3_' + np.min.__name__ + '_dif'] = X[feature12] - X[feature12 + '_action3_' + np.min.__name__ ]
X[feature23 + '_action1_' + np.min.__name__ + '_dif'] = X[feature23] - X[feature23 + '_action1_' + np.min.__name__ ]
X[feature23 + '_action2_' + np.min.__name__ + '_dif'] = X[feature23] - X[feature23 + '_action2_' + np.min.__name__ ]
X[feature23 + '_action3_' + np.min.__name__ + '_dif'] = X[feature23] - X[feature23 + '_action3_' + np.min.__name__ ]
X[feature13 + '_action1_' + np.min.__name__ + '_dif'] = X[feature13] - X[feature13 + '_action1_' + np.min.__name__ ]
X[feature13 + '_action2_' + np.min.__name__ + '_dif'] = X[feature13] - X[feature13 + '_action2_' + np.min.__name__ ]
X[feature13 + '_action3_' + np.min.__name__ + '_dif'] = X[feature13] - X[feature13 + '_action3_' + np.min.__name__ ]
end_ind_col = len(X.columns)
return X, list(range(start_ind_col, end_ind_col))
def _static_feature_generation(X):
X = _generate_pref_gaps(X)
X = _generate_gaps(X)
#X = _generate_scenario_type(X)
return X
def _dynamic_feature_generation(X, X_train, y_train):
all_voters = pd.DataFrame(X[["VoterID", "SessionIDX"]].drop_duplicates())
for voter in all_voters.iterrows():
X, added_columns = _generate_A_ratios(X, X_train, y_train, voter[1])
if len(added_columns)>0:
a_ratio_columns = added_columns
X, added_columns = _generate_gaps_features(X, X_train, y_train, voter[1])
if len(added_columns)>0:
gaps_columns = added_columns
X = _generate_action_aggregation_features(X, X_train, y_train, voter[1])
#Gaps features encoding
X = X.fillna(X.mean()) #X.fillna(1000) #fill na with some high value (maybe maximum) because the voters with na values didn't choose the action (say q'', 3) in all gaps they incounterd.
X, gaps_dif_columns = _generate_gap_dif_features(X)
total_gaps_columns = a_ratio_columns + gaps_columns + gaps_dif_columns
total_gaps_columns.append(X.columns.get_loc("GAP12_pref_poll"))
total_gaps_columns.append(X.columns.get_loc("GAP23_pref_poll"))
total_gaps_columns.append(X.columns.get_loc("GAP13_pref_poll"))
normalized_gap_fs = pd.DataFrame(preprocessing.normalize(X.iloc[:,total_gaps_columns]))
encoded_gap_fs = pd.DataFrame(_autoencode(normalized_gap_fs))
X = pd.concat([X, encoded_gap_fs], axis=1, join='inner')
X = X.drop(X.columns[gaps_columns + gaps_dif_columns], axis=1)
X = _generate_is_random_voter(X)
X = _generate_voter_type(X)
return X
def _evaluation(X, clf, target):
#tests
#static features generation
X = _static_feature_generation(X)
# Encoders definitions
le = sklearn.preprocessing.LabelEncoder()
target_le = sklearn.preprocessing.LabelEncoder()
# Split into features and target
features_df, target_df = X.drop([target], axis=1),X[target]
n_folds = 10
results_df = pd.DataFrame(columns=['Measure', 'Result'])
# Initialize metrics:
results_df.loc[0] = ['PRECISION', 0]
results_df.loc[1] = ['RECALL', 0]
results_df.loc[2] = ['F_MEASURE', 0]
# 10 fold cross validation
kf = KFold(n_folds, shuffle=True, random_state=1) # 10 fold cross validation
for train_indices, test_indices in kf.split(features_df, target_df):
# Feature Generation
features_train = features_df.loc[[ii for ii in train_indices],]
targets_train = target_df[[ii for ii in train_indices]]
features_ext_df = _dynamic_feature_generation(features_df, features_train, targets_train)
features_ext_df = features_ext_df.drop(["Action_name"], axis=1)
# encoding the dataframes
features_encoded_df = pd.DataFrame(preprocessing.normalize(preprocessing.scale(_data_cleaning(features_ext_df, False, le))))
target_encoded_df = _data_cleaning(target_df, True, target_le)
# make training and testing datasets
features_train = features_encoded_df.loc[[ii for ii in train_indices],]
features_test = features_encoded_df.loc[[ii for ii in test_indices],]
targets_train = target_encoded_df[[ii for ii in train_indices]]
targets_test = target_encoded_df[[ii for ii in test_indices]]
# Train
clf.fit(X = features_train.as_matrix(), y = targets_train)
# Test
predicated = clf.predict(features_test.as_matrix())
# Measures
results_df.iloc[0, 1] = results_df.iloc[0, 1] + precision_score(targets_test, predicated, average='weighted')
results_df.iloc[1, 1] = results_df.iloc[1, 1] + recall_score(targets_test, predicated, average='weighted')
results_df.iloc[2, 1] = results_df.iloc[2, 1] + f1_score(targets_test, predicated, average='weighted')
results_df.Result = results_df.Result.apply(lambda x: x / n_folds)
return results_df
def intersection(lst1, lst2):
lst3 = [value for value in lst1 if value in lst2]
return lst3
def _read_roy_folds(folds_file):
lines = folds_file.read().split('\n')
folds = list()
for index in range(0,len(lines)):
line = lines[index]
if not ('fold' in line) and line != '':
folds.append([int(ii) for ii in line[1:len(line)-1].split(',')])
return folds
def _get_loo_folds(X):
folds = list()
for x in X.iterrows():
fold = [x[0]]
folds.append(fold)
return folds
def _autoencode(features):
#test
encoding_dim = int(len(features.columns)/5)
input_votes = Input(shape = (len(features.columns),))
encoded = Dense(encoding_dim, activation='relu')(input_votes)
decoded = Dense(len(features.columns), activation='tanh')(encoded)
autoencoder = Model(input_votes, decoded)
encoder = Model(input_votes, encoded)
# encoded_input = Input(shape=(encoding_dim,))
# decoder_layer = autoencoder.layers[-1]
# decoder = Model(encoded_input, decoder_layer(encoded_input))
autoencoder.compile(optimizer='adadelta', loss='MSE')
autoencoder.fit(features, features,
epochs=150,
batch_size=256,
shuffle=True,verbose=False)
encoded_votes = encoder.predict(features)
return encoded_votes
def _evaluation_roy_splits(raw_data, clfs, target, folds, scenario_filter):
data = raw_data.copy()
#static features generation
data = _static_feature_generation(data)
# Encoders definitions
le = sklearn.preprocessing.LabelEncoder()
target_le = sklearn.preprocessing.LabelEncoder()
n_folds = 10
results_df = pd.DataFrame(columns=['Classifier', 'PRECISION', 'RECALL','F_MEASURE','ACCURACY'])
prediction = pd.DataFrame(np.matrix([]))
feature_importances = pd.DataFrame()
features_train = pd.DataFrame()
# 10 fold cross validation
for i in range(0,len(folds)):
# Split into features and target
features_df, target_df = data.drop([target], axis=1),data[target]
test_indices = data.index[[x[1].RoundIndex in folds[i] for x in data.iterrows()]].tolist()
train_indices = data.index[[not (x[1].RoundIndex in folds[i] or x[1].Scenario == scenario_filter) for x in data.iterrows()]].tolist()
# Feature Generation
features_train = features_df.loc[[ii for ii in train_indices],]
targets_train = target_df[[ii for ii in train_indices]]
features_ext_df = _dynamic_feature_generation(features_df, features_train , targets_train)
features_ext_df = features_ext_df.drop(["Action_name"], axis=1)
features_ext_df = features_ext_df.drop(["Vote"], axis=1)
# encoding the dataframes
features_encoded_df = pd.DataFrame(preprocessing.normalize(preprocessing.scale(_data_cleaning(features_ext_df, False, le).as_matrix()), axis=0, norm='max'))#
target_encoded_df = target_df #_data_cleaning(target_df, True, target_le)
# make training and testing datasets
features_train = features_encoded_df.loc[[ii for ii in train_indices],]
features_test = features_encoded_df.loc[[ii for ii in test_indices],]
targets_train = target_encoded_df[[ii for ii in train_indices]]
targets_test = target_encoded_df[[ii for ii in test_indices]]
# training = pd.concat([features_train, pd.DataFrame(targets_train)], axis=1, join='inner')
# testing = pd.concat([features_test, pd.DataFrame(targets_test)], axis=1, join='inner')
# training.to_csv("training_fold_"+str(i)+".csv")
# testing.to_csv("testing_fold_"+str(i)+".csv")
for j in range(0,len(clfs)):
clf = clfs[j]
clf_name = str(clf).split("(")[0]
if i == 0:
#Initialize metrics
results_df.loc[j] = [str(clf), 0, 0, 0, 0]
if clf != "baseline":
# Train
clf.fit(X = features_train.as_matrix(), y = targets_train)
# Test
predicated = clf.predict(features_test.as_matrix())
# #feature importance
# current_feature_importances = pd.DataFrame(clf.feature_importances_,
# index=features_ext_df.columns,
# columns=['importance']).sort_values('importance',
# ascending=False)
# if len(feature_importances) == 0:
# feature_importances = current_feature_importances
# else:
# feature_importances['importance'] = feature_importances['importance'] + current_feature_importances['importance']
#
# print(feature_importances)
else:
features_ext_df_test = features_ext_df.loc[[ii for ii in test_indices],]
n_samples = len(features_ext_df_test)
predicated = np.zeros((n_samples),dtype=int)
features_ext_df_test["Prediction"] = 1
features_ext_df_test.loc[(features_ext_df_test["Scenario"] == "C") & (features_ext_df_test["VoterType"] == "LB"), "Prediction"] = 2
features_ext_df_test.loc[(features_ext_df_test['Scenario'].isin(["E","F"])) & (features_ext_df_test["VoterType"] != "TRT"), "Prediction"] = 2
predicated = features_ext_df_test["Prediction"]
features_ext_df_test = pd.concat([features_ext_df_test, pd.DataFrame(predicated)], axis=1)
#aggregate results
if len(prediction) == 0:
prediction = pd.DataFrame(predicated)
else:
prediction = pd.concat([prediction, pd.DataFrame(predicated)])
raw_data.loc[[ii for ii in test_indices],"Prediction" + "_" + clf_name] = predicated
print(str(clf) +": F_score = " + str(f1_score(targets_test, predicated, average='weighted')))
# Measures
results_df.iloc[j, 1] = results_df.iloc[j, 1] + precision_score(targets_test, predicated, average='weighted')
results_df.iloc[j, 2] = results_df.iloc[j, 2] + recall_score(targets_test, predicated, average='weighted')
results_df.iloc[j, 3] = results_df.iloc[j, 3] + f1_score(targets_test, predicated, average='weighted')
results_df.iloc[j, 4] = results_df.iloc[j, 4] + accuracy_score(targets_test, predicated)
results_df.Result = results_df.Result.apply(lambda x: x / n_folds)
return results_df, prediction, raw_data, feature_importances, features_train
def _build_data_by_folds(data, folds):
transformed_data = pd.DataFrame()
for i in range(0,len(folds)):
# Split into features and target
fold_indices = data.index[[x[1].RoundIndex in folds[i] for x in data.iterrows()]].tolist()
fold_df = data.iloc[fold_indices,:]
if len(transformed_data) == 0:
transformed_data = fold_df
else:
transformed_data = pd.concat([transformed_data, fold_df])
return transformed_data
#
# data = pd.read_excel("datasets/oneshot/full_data_no_timeout_no_first_round.xlsx")
#
# data_with_action_name = _generate_action_name(data)
# data_with_action_name.to_csv("datasets\\oneshot\\full_data_no_timeout_no_first_round_with_action_name.csv")
#
# #Load rawdata
# oneshot_df = pd.read_csv("datasets/oneshot/full_data_no_timeout_no_first_round_with_action_name.csv")#read_excel("datasets/oneshot/NewData.xlsx")
#
# #sorted_data = _build_data_by_folds(oneshot_df)
# #sorted_data.to_csv("Results\\sorted_fulldata.csv")
# #
# #print("F_score = " + str(f1_score(sorted_data.Action, nn_prediction[1:], average='weighted')))
# #
#
# raw_data = oneshot_df
#
#
#
# raw_data = raw_data.drop(['GameConfiguration','FileName','PlayerID','DateTime','CandName1', 'CandName2','CandName3','WinnerPostVote', 'VotesCand1', 'VotesCand2', 'VotesCan3', 'PointsPostVote','ResponseTime'], axis=1)
classifier = RandomForestClassifier(n_estimators=100)#MLPClassifier(hidden_layer_sizes = (92), max_iter = 500)
#
# #Model Phase
# #rf_results_df = _evaluation_roy_splits(data, RandomForestClassifier(n_estimators=100), 'Action')
#
#
# nn_results_df, nn_prediction, data_with_pred = _evaluation_roy_splits(raw_data, classifier, 'Action')
#
# nn_results_df.to_csv("Results\\rf_new_performance_df.csv")
# nn_prediction.to_csv("Results\\rf_new_prediction.csv")
# data_with_pred.to_csv("Results\\rf_new_data_with_pred.csv")
#
#baseline_results_df, baseline_prediction, baseline_full_pred = _evaluation_roy_splits(data, "baseline", 'Action')
#
#
#baseline_results_df.to_csv("Results\\baseline_performance_df.csv")
#baseline_prediction.to_csv("Results\\baseline_prediction.csv")
#baseline_full_pred.to_csv("Results\\baseline_full_pred.csv")
#
# # #other data sets
# data = pd.read_excel("datasets/oneshot/tal.xlsx")
# data_with_action_name = _generate_action_name(data)
# data_with_action_name.to_csv("datasets\\oneshot\\tal_no_timeout_with_action_name.csv")
# folds = _read_roy_folds(open("datasets/oneshot/tal_folds.txt", "r"))
# #
#
# oneshot_df = pd.read_csv("datasets/oneshot/tal_no_timeout_with_action_name.csv")#read_excel("datasets/oneshot/NewData.xlsx")
# oneshot_df = oneshot_df.fillna(oneshot_df.mean())
#
#
# raw_data = oneshot_df
#
# raw_data = raw_data.drop(['DateTime','Util4','LabExperiment','ExperimentComments','CandName1', 'CandName2','CandName3','CanName4','GameID', 'PrefsObserved','Pref4', 'VotesCand1','ResponseTime', 'VotesCand2', 'VotesCand3'], axis=1)
#
# #Model Phase
# #rf_results_df = _evaluation_roy_splits(data, RandomForestClassifier(n_estimators=100), 'Action')
#
#
# tal_nn_performance_df, tal_nn_prediction, tal_nn_pred = _evaluation_roy_splits(raw_data, classifier, 'Action', folds)
#
# tal_nn_performance_df.to_csv("Results\\tal_rf_performance_df.csv")
# tal_nn_prediction.to_csv("Results\\tal_rf_prediction.csv")
# tal_nn_pred.to_csv("Results\\tal_rf_pred.csv")
#
#
# print("F_score = " + str(f1_score(tal_nn_pred.Action, tal_nn_pred.nn_predction.astype(int), average='weighted')))
#
# #scharm
# scharm_data = pd.read_excel("datasets/oneshot/schram.xlsx")
#
# data_with_action_name = _generate_action_name(scharm_data)
# data_with_action_name.to_csv("datasets\\oneshot\\scharm_data_with_action_name.csv")
# #other data sets
# scharm_df = pd.read_csv("datasets/oneshot/scharm_data_with_action_name.csv")#read_excel("datasets/oneshot/NewData.xlsx")
# scharm_df = scharm_df.fillna(scharm_df.mean())
# folds = _read_roy_folds(open("datasets/oneshot/schram_folds.txt", "r"))
#
# raw_data = scharm_df
#
# #raw_data = raw_data.drop(['DateTime','Util4','LabExperiment','ExperimentComments','CandName1', 'CandName2','CandName3','CanName4','GameID', 'PrefsObserved','Pref4', 'VotesCand1','ResponseTime', 'VotesCand2', 'VotesCand3'], axis=1)
#
# #Model Phase
# #rf_results_df = _evaluation_roy_splits(data, RandomForestClassifier(n_estimators=100), 'Action')
#
#
# scharm_nn_performance_df, scharm_nn_prediction, scharm_nn_pred = _evaluation_roy_splits(raw_data, classifier, 'Action', folds)
#
# scharm_nn_performance_df.to_csv("Results\\scharm_rf_performance_df.csv")
# scharm_nn_prediction.to_csv("Results\\scharm_rf_prediction.csv")
# scharm_nn_pred.to_csv("Results\\scharm_rf_pred.csv")
#
# print("F_score = " + str(f1_score(scharm_nn_pred.Action, scharm_nn_pred.nn_predction.astype(int), average='weighted')))
# # d32
# d32_data = pd.read_excel("datasets/oneshot/d32_updated.xlsx")
#
#
# data_with_action_name = _generate_action_name(d32_data)
# data_with_action_name.to_csv("datasets\\oneshot\\d32_data_with_action_name.csv")
# #other data sets
d32_df = pd.read_csv("datasets/oneshot/d32_data_with_action_name.csv")#read_excel("datasets/oneshot/NewData.xlsx")
d32_df = d32_df.fillna(d32_df.mean())
folds = _read_roy_folds(open("datasets/oneshot/d32_folds.txt", "r"))
#Model Phase
#rf_results_df = _evaluation_roy_splits(data, RandomForestClassifier(n_estimators=100), 'Action')
for scenario in ['NONE']:#['A','B','C','D','E','F','NONE']:
raw_data = d32_df.copy()
raw_data = raw_data.drop(
['DateTime', 'CandName1', 'CandName2', 'CandName3', 'VotesCand1', 'ResponseTime', 'VotesCand2', 'VotesCan3',
'PointsPostVote', 'WinnerPostVote', 'AU', 'KP', 'LD', 'LDLB', 'CV', 'BW', 'HF', 'heuristic', 'parameter'],
axis=1)
# loo_folds = _get_loo_folds(raw_data)
personal_rf_clf = PersonalClassifier(id_index=raw_data.columns.get_loc("VoterID"), n_upsample=3)#RandomForestClassifier(n_estimators=100) # MLPClassifier(hidden_layer_sizes = (92), max_iter = 500)
personal_nn_clf = PersonalClassifier(id_index=raw_data.columns.get_loc("VoterID"), base_classifier=MLPClassifier(hidden_layer_sizes = (92), max_iter = 500), n_upsample=10, general_base_classifier=True)#RandomForestClassifier(n_estimators=100) # MLPClassifier(hidden_layer_sizes = (92), max_iter = 500)
neural_net_cf = MLPClassifier(hidden_layer_sizes = (92), max_iter = 500)
rf_clf = RandomForestClassifier(n_estimators=100)
dt_clf = DecisionTreeClassifier()
adaboost_clf = AdaBoostClassifier(n_estimators=200)
svm_clf = SVC()
logistics_clf = logistic.LogisticRegression()
classifiers = [personal_rf_clf, personal_nn_clf, neural_net_cf, rf_clf, dt_clf, adaboost_clf, svm_clf, logistics_clf]
d32_performance_df,d32_prediction, d32_pred, feature_importances, features_train = _evaluation_roy_splits(raw_data, classifiers, 'Action', folds, scenario)
d32_performance_df.to_csv("Results\\d32_performance_df_"+scenario+".csv")
# d32_prediction.to_csv("Results\\d32_rf_prediction_"+scenario+".csv")
d32_pred.to_csv("Results\\d32_pred_"+scenario+".csv")
# feature_importances.to_csv("Results\\d32_feature_importance_"+scenario+".csv")
# features_train.to_csv("Results\\d32_feature_train_"+scenario+".csv")
print("F_score = " + str(f1_score(d32_pred.Action, d32_pred.Prediction.astype(int), average='weighted')))
#d36
# d36_data = pd.read_excel("datasets/oneshot/d36_updated.xlsx")
#
# data_with_action_name = _generate_action_name(d36_data)
# data_with_action_name.to_csv("datasets\\oneshot\\d36_data_with_action_name.csv")
# #other data sets
# d36_df = pd.read_csv("datasets/oneshot/d36_data_with_action_name.csv")#read_excel("datasets/oneshot/NewData.xlsx")
# d36_df = d36_df.fillna(d36_df.mean())
# folds = _read_roy_folds(open("datasets/oneshot/d36_folds.txt", "r"))
#
#
#
# #Model Phase
# #rf_results_df = _evaluation_roy_splits(data, RandomForestClassifier(n_estimators=100), 'Action')
#
# for scenario in ['A','B','C','D','E','F','NONE']:
# classifier = RandomForestClassifier(n_estimators=100) # MLPClassifier(hidden_layer_sizes = (92), max_iter = 500)
# raw_data = d36_df.copy()
#
# raw_data = raw_data.drop(['DateTime', 'CandName1', 'CandName2', 'CandName3', 'VotesCand1', 'ResponseTime', 'VotesCand2', 'VotesCan3',
# 'PointsPostVote', 'WinnerPostVote', 'AU', 'KP', 'LD', 'LDLB', 'CV', 'BW', 'HF', 'heuristic', 'parameter'],
# axis=1)
#
# d36_nn_performance_df,d36_nn_prediction, d36_nn_pred,feature_importances,features_train = _evaluation_roy_splits(raw_data, classifier, 'Action', folds,scenario)
#
# d36_nn_performance_df.to_csv("Results\\d36_rf_performance_df_"+scenario+".csv")
# d36_nn_prediction.to_csv("Results\\d36_rf_prediction_"+scenario+".csv")
# d36_nn_pred.to_csv("Results\\d36_rf_pred_"+scenario+".csv")
# feature_importances.to_csv("Results\\d36_feature_importance_"+scenario+".csv")
# features_train.to_csv("Results\\d36_feature_train_"+scenario+".csv")
#
# d36_nn_pred = _convert_prediction(d36_nn_pred)
#
# d36_nn_pred.to_csv("Results\\V_d36_rf_pred_"+scenario+".csv")
#
#
# print("F_score = " + str(f1_score(d36_nn_pred.Action, d36_nn_pred.Prediction.astype(int), average='weighted')))
#
#