-
Notifications
You must be signed in to change notification settings - Fork 5
/
Copy pathbinary.py
1084 lines (914 loc) · 53.2 KB
/
binary.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
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
# -*- coding: utf-8 -*-
"""
Created on Fri Jun 21 16:37:09 2019
@author: zhangli
"""
import numpy as np
import pandas as pd
import lightgbm as lgb
import os
import seaborn as sns
import matplotlib.pyplot as plt
import warnings
warnings.filterwarnings("ignore")
from tqdm import tqdm
import json
from sklearn.metrics import f1_score
from sklearn.model_selection import StratifiedKFold
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.decomposition import TruncatedSVD
from itertools import product
import ast
import math
from sklearn.cluster import KMeans
import time
from datetime import datetime
import gc
######################################## 基本数据集加载 ########################################
data_path = '../data_set_phase2/data/'
print('基本数据集加载')
data = pd.read_csv(data_path + 'shanghai.csv',parse_dates=['req_time','plan_time'])
profiles = pd.read_csv('../data_set_phase2/data/profiles.csv')
print('完成加载')
########################################## 特征工程 ###########################################
print('特征工程')
######################################## 经纬度特征 ##########################################
print('经纬度特征')
#1、曼哈顿距离特征：两点投影到各轴上的距离总和
data['o_lng'] = data['o'].apply(lambda x: float(x.split(',')[0]))
data['o_lat'] = data['o'].apply(lambda x: float(x.split(',')[1]))
data['d_lng'] = data['d'].apply(lambda x: float(x.split(',')[0]))
data['d_lat'] = data['d'].apply(lambda x: float(x.split(',')[1]))
data['od_manhattan_distance'] = abs(data['o_lng']-data['d_lng'])+abs(data['o_lat']-data['d_lat'])
#2、直线距离特征：根据经纬度坐标计算两点间的直线距离
def GetDistance(lng1, lat1, lng2, lat2):
#地球赤道半径6378.137km
EARTH_RADIUS = 6378.137
#把经纬度转换成度（°）
lng1 = lng1 * (math.pi / 180.0)
lng2 = lng2 * (math.pi / 180.0)
lat1 = lat1 * (math.pi / 180.0)
lat2 = lat2 * (math.pi / 180.0)
dis1 = lat1 - lat2
dis2 = lng1 - lng2
#以1m为球的半径，求球上两点的距离
s = 2 * math.asin(
((math.sin(dis1 / 2)) ** 2 + math.cos(lat1) * math.cos(lat2) * (math.sin(dis2 / 2)) ** 2) ** 0.5)
#换算成地球的半径：6378.137km *1000 = 6378137m
s = s * EARTH_RADIUS * 1000
return s
data['od_manhattan_distance_detail']=data.apply(lambda row:GetDistance(row['o_lng'],row['o_lat'],row['d_lng'],row['d_lat']),axis=1)
#3、始终点的经纬度分别与始终点的经纬度的平均值和众数的差之和
o_co = data[['o']]
d_co = data[['d']]
o_co.columns = ['co']
d_co.columns = ['co']
all_co = pd.concat([d_co, o_co]).drop_duplicates()
all_co['lng'] = all_co['co'].apply(lambda x: float(x.split(',')[0]))
all_co['lat'] = all_co['co'].apply(lambda x: float(x.split(',')[1]))
lng_mean = all_co['lng'].mean()
lat_mean = all_co['lat'].mean()
lng_mode = all_co['lng'].mode()[0]
lat_mode = all_co['lat'].mode()[0]
data['o_main_centroid_mean_dis'] = abs(
data['o_lng']-lng_mean)+abs(data['o_lat']-lat_mean)
data['d_main_centroid_mean_dis'] = abs(
data['d_lng']-lng_mean)+abs(data['d_lat']-lat_mean)
data['o_main_centroid_mode_dis'] = abs(
data['o_lng']-lng_mode)+abs(data['o_lat']-lat_mode)
data['d_main_centroid_mode_dis'] = abs(
data['d_lng']-lng_mode)+abs(data['d_lat']-lat_mode)
#4、所有地点的经纬度聚类
data['o_cluster', 'd_cluster'] = np.nan
clf_KMeans = KMeans(n_clusters=11)#构造聚类器
cluster = clf_KMeans.fit_predict(all_co[['lng','lat']])#聚类
index = 0
for co in tqdm(all_co['co']):
#所有地点的经纬度聚类
data.loc[(data['o'] == co), 'o_cluster'] = cluster[index]
data.loc[(data['d'] == co), 'd_cluster'] = cluster[index]
index +=1
od_feature = ['od_manhattan_distance', 'od_manhattan_distance_detail', 'o_main_centroid_mean_dis', 'd_main_centroid_mean_dis', 'o_main_centroid_mode_dis', 'd_main_centroid_mode_dis', 'o_cluster', 'd_cluster']
print('经纬度特征数量：', len(od_feature), od_feature)
######################################## 距离特征 ##########################################
print('距离特征')
def minkowski_distance(x1, x2, y1, y2, p):
return ((abs(x2 - x1) ** p) + (abs(y2 - y1)) ** p) ** (1.0 / p)
def calculate_direction(d_lon, d_lat):
result = np.zeros(len(d_lon))
l = np.sqrt(d_lon**2 + d_lat**2)
result[d_lon>0] = (180/np.pi)*np.arcsin(d_lat[d_lon>0]/l[d_lon>0])
idx = (d_lon<0) & (d_lat>0)
result[idx] = 180 - (180/np.pi)*np.arcsin(d_lat[idx]/l[idx])
idx = (d_lon<0) & (d_lat<0)
result[idx] = -180 - (180/np.pi)*np.arcsin(d_lat[idx]/l[idx])
return result
def add_travel_vector_features(df):
df['delta_longitude'] = df.o_lng - df.d_lng
df['delta_latitude'] = df.o_lat - df.d_lat
df['pickup_x'] = np.cos(df.o_lat) * np.cos(df.o_lng)
df['pickup_y'] = np.cos(df.o_lat) * np.sin(df.o_lng)
df['pickup_z'] = np.sin(df.o_lat)
df['dropoff_x'] = np.cos(df.d_lat) * np.cos(df.d_lng)
df['dropoff_y'] = np.cos(df.d_lat) * np.sin(df.d_lng)
df['dropoff_z'] = np.sin(df.d_lat)
data['manhattan'] = minkowski_distance(data.o_lng, data.d_lng,
data.o_lat, data.d_lat, 1)
data['euclidean']=minkowski_distance(data.o_lng, data.d_lng,
data.o_lat, data.d_lat, 2)
add_travel_vector_features(data)
data['direction'] = calculate_direction(data.delta_longitude, data.delta_latitude)
distance_feature = ['manhattan','euclidean','delta_longitude','delta_latitude','pickup_x','pickup_y','pickup_z','dropoff_x','dropoff_y','dropoff_z','direction']
print('距离特征数量：', len(distance_feature), distance_feature)
################################ plan_time & req_time特征 ###################################
print('plan_time & req_time特征')
#1、原始时间特征上的处理
data['req_time_hour'] = data['req_time'].dt.hour
data['req_time_weekday'] = data['req_time'].dt.weekday
data['req_time_minute'] = data['req_time'].dt.minute
data['req_time_date_d'] = data['req_time'].dt.strftime('%d').astype(int)
#2、时差特征
data['time_diff'] = (data['plan_time']- data['req_time']).dt.seconds
#3、请求的时间点与当天早上6点、中午12点、下午6点、晚上6点的分钟差距
def diff_6_clock(time_point):
six_clock=str(time_point.year)+'-'+str(time_point.month)+'-'+str(time_point.day)+' '+str(6)+':0:00'
six_clock=pd.to_datetime(six_clock)
the_diff=(six_clock- time_point).total_seconds()
return abs(the_diff/60)
def diff_12_clock(time_point):
t12_clock=str(time_point.year)+'-'+str(time_point.month)+'-'+str(time_point.day)+' '+str(12)+':0:00'
t12_clock=pd.to_datetime(t12_clock)
the_diff=(t12_clock- time_point).total_seconds()
return abs(the_diff/60)
def diff_18_clock(time_point):
t18_clock=str(time_point.year)+'-'+str(time_point.month)+'-'+str(time_point.day)+' '+str(18)+':0:00'
t18_clock=pd.to_datetime(t18_clock)
the_diff=(t18_clock- time_point).total_seconds()
return abs(the_diff/60)
def diff_24_clock(time_point):
t24_clock=str(time_point.year)+'-'+str(time_point.month)+'-'+str(time_point.day)+' '+str(23)+':59:00'
t24_clock=pd.to_datetime(t24_clock)
the_diff=(t24_clock- time_point).total_seconds()
return abs(the_diff/60)
data['diff_6_cloc']=data['req_time'].apply(diff_6_clock)
data['diff_12_clock']=data['req_time'].apply(diff_12_clock)
data['diff_18_clock']=data['req_time'].apply(diff_18_clock)
data['diff_24_clock']=data['req_time'].apply(diff_24_clock)
time_feature = ['req_time_hour', 'req_time_weekday', 'req_time_minute', 'req_time_date_d', 'time_diff',
'diff_6_cloc', 'diff_12_clock', 'diff_18_clock', 'diff_24_clock']
print('plan_time & req_time特征数量：', len(time_feature),time_feature)
################################ 对profile的降维提取特征 ###################################
print('对profile的降维提取特征')
x = profiles.drop(['pid'], axis=1).values
svd = TruncatedSVD(n_components=10, n_iter=20, random_state=2019)
svd_x = svd.fit_transform(x)
svd_feas = pd.DataFrame(svd_x)
svd_feas.columns = ['svd_fea_{}'.format(i) for i in range(10)]
svd_feas['pid'] = profiles['pid'].values
data = data.merge(svd_feas, on='pid', how='left')
profiles_svd_feature = ['svd_fea_{}'.format(i) for i in range(10)]
print('对profile的降维提取特征完成数量：', len(profiles_svd_feature), profiles_svd_feature)
############################## plans百度地图推荐交通方式特征 #################################
# ### plans百度地图推荐特征
#
# 此处特征包括：
# *Plans中针对这次导航任务，百度地图所提供的路线选择:12
# *提供的路线选择中距离、价格和所花时间的最大值、最小值、均值、方差:12
# *提供的路线选择中距离最大和最小、价格最高和最低、所花时间最多和最少的交通方式:6
# *提供的路线选择中的第一种交通方式:1
# *模式数据压缩:10
print('plans百度地图推荐交通方式特征')
data['plans_json'] = data['plans'].fillna('[]').apply(lambda x: json.loads(x))
def gen_plan_feas(data):
n = data.shape[0]
mode_list_feas = np.zeros((n, 12))
max_dist, min_dist, mean_dist, std_dist = np.zeros((n,)), np.zeros((n,)), np.zeros((n,)), np.zeros((n,))
max_price, min_price, mean_price, std_price = np.zeros((n,)), np.zeros((n,)), np.zeros((n,)), np.zeros((n,))
max_eta, min_eta, mean_eta, std_eta = np.zeros((n,)), np.zeros((n,)), np.zeros((n,)), np.zeros((n,))
min_dist_mode, max_dist_mode, min_price_mode, max_price_mode, min_eta_mode, max_eta_mode, first_mode = np.zeros((n,)), np.zeros((n,)), np.zeros((n,)), np.zeros((n,)), np.zeros((n,)), np.zeros((n,)), np.zeros((n,))
mode_texts = []
for i, plan in tqdm(enumerate(data['plans_json'].values)):
if len(plan) == 0:
cur_plan_list = []
else:
cur_plan_list = plan
if len(cur_plan_list) == 0:
mode_list_feas[i, 0] = 1
first_mode[i] = 0
max_dist[i] = -1
min_dist[i] = -1
mean_dist[i] = -1
std_dist[i] = -1
max_price[i] = -1
min_price[i] = -1
mean_price[i] = -1
std_price[i] = -1
max_eta[i] = -1
min_eta[i] = -1
mean_eta[i] = -1
std_eta[i] = -1
min_dist_mode[i] = -1
max_dist_mode[i] = -1
min_price_mode[i] = -1
max_price_mode[i] = -1
min_eta_mode[i] = -1
max_eta_mode[i] = -1
mode_texts.append('word_null')
else:
distance_list = []
price_list = []
eta_list = []
mode_list = []
for tmp_dit in cur_plan_list:
distance_list.append(int(tmp_dit['distance']))
if tmp_dit['price'] == '':
price_list.append(0)
else:
price_list.append(int(tmp_dit['price']))
eta_list.append(int(tmp_dit['eta']))
mode_list.append(int(tmp_dit['transport_mode']))
mode_texts.append(
' '.join(['word_{}'.format(mode) for mode in mode_list]))
distance_list = np.array(distance_list)
price_list = np.array(price_list)
eta_list = np.array(eta_list)
mode_list = np.array(mode_list, dtype='int')
mode_list_feas[i, mode_list] = 1
distance_sort_idx = np.argsort(distance_list)
price_sort_idx = np.argsort(price_list)
eta_sort_idx = np.argsort(eta_list)
max_dist[i] = distance_list[distance_sort_idx[-1]]
min_dist[i] = distance_list[distance_sort_idx[0]]
mean_dist[i] = np.mean(distance_list)
std_dist[i] = np.std(distance_list)
max_price[i] = price_list[price_sort_idx[-1]]
min_price[i] = price_list[price_sort_idx[0]]
mean_price[i] = np.mean(price_list)
std_price[i] = np.std(price_list)
max_eta[i] = eta_list[eta_sort_idx[-1]]
min_eta[i] = eta_list[eta_sort_idx[0]]
mean_eta[i] = np.mean(eta_list)
std_eta[i] = np.std(eta_list)
first_mode[i] = mode_list[0]
max_dist_mode[i] = mode_list[distance_sort_idx[-1]]
min_dist_mode[i] = mode_list[distance_sort_idx[0]]
max_price_mode[i] = mode_list[price_sort_idx[-1]]
min_price_mode[i] = mode_list[price_sort_idx[0]]
max_eta_mode[i] = mode_list[eta_sort_idx[-1]]
min_eta_mode[i] = mode_list[eta_sort_idx[0]]
feature_data = pd.DataFrame(mode_list_feas)
feature_data.columns = ['mode_feas_{}'.format(i) for i in range(12)]
feature_data['max_dist'] = max_dist
feature_data['min_dist'] = min_dist
feature_data['mean_dist'] = mean_dist
feature_data['std_dist'] = std_dist
feature_data['max_price'] = max_price
feature_data['min_price'] = min_price
feature_data['mean_price'] = mean_price
feature_data['std_price'] = std_price
feature_data['max_eta'] = max_eta
feature_data['min_eta'] = min_eta
feature_data['mean_eta'] = mean_eta
feature_data['std_eta'] = std_eta
feature_data['max_dist_mode'] = max_dist_mode
feature_data['min_dist_mode'] = min_dist_mode
feature_data['max_price_mode'] = max_price_mode
feature_data['min_price_mode'] = min_price_mode
feature_data['max_eta_mode'] = max_eta_mode
feature_data['min_eta_mode'] = min_eta_mode
feature_data['first_mode'] = first_mode
print('mode tfidf...')
tfidf_enc = TfidfVectorizer(ngram_range=(1, 2))
tfidf_vec = tfidf_enc.fit_transform(mode_texts)
svd_enc = TruncatedSVD(n_components=10, n_iter=20, random_state=2019)
mode_svd = svd_enc.fit_transform(tfidf_vec)
mode_svd = pd.DataFrame(mode_svd)
mode_svd.columns = ['svd_mode_{}'.format(i) for i in range(10)]
plan_fea = pd.concat([feature_data, mode_svd], axis=1)
plan_fea['sid'] = data['sid'].values
return plan_fea
data_plans = gen_plan_feas(data)
data = data.merge(data_plans, on='sid', how='left')
plan_feature = ['mode_feas_0', 'mode_feas_1', 'mode_feas_2', 'mode_feas_3', 'mode_feas_4', 'mode_feas_5', 'mode_feas_6', 'mode_feas_7', 'mode_feas_8', 'mode_feas_9', 'mode_feas_10', 'mode_feas_11',
'max_dist', 'min_dist', 'mean_dist', 'std_dist', 'max_price', 'min_price', 'mean_price', 'std_price', 'max_eta', 'min_eta', 'mean_eta', 'std_eta', 'max_dist_mode', 'min_dist_mode', 'max_price_mode', 'min_price_mode', 'max_eta_mode', 'min_eta_mode',
'first_mode', 'svd_mode_0', 'svd_mode_1', 'svd_mode_2', 'svd_mode_3', 'svd_mode_4', 'svd_mode_5', 'svd_mode_6', 'svd_mode_7', 'svd_mode_8', 'svd_mode_9']
print('plans百度地图推荐交通方式特征数量：', len(plan_feature), plan_feature)
############################## plan推荐mode平铺特征 #################################
print('plan推荐mode平铺特征')
#对plans内的信息直接进行平铺，平铺时求取每个mode的距离、price、eta，根据mode在plan中出现的次序给出排名信息。这样就构建了11*4个特征
data['plans_json'] = data['plans'].fillna('[]').apply(lambda x: json.loads(x))
def gen_plan_pingpu_feas(data):
#创建用于放置mode平均距离、平均价格、平均eta、出现位置排名,共11个模式的11*4个特征 （为0的模式就不做标识了） 添加model所属的工具模态
n = data.shape[0]
#mode_list_feas = np.zeros((n, 12))
model_1_dist,model_1_price,model_1_eta,model_1_rank = np.zeros((n,)), np.zeros((n,)), np.zeros((n,)), np.zeros((n,))
model_2_dist,model_2_price,model_2_eta,model_2_rank = np.zeros((n,)), np.zeros((n,)), np.zeros((n,)), np.zeros((n,))
model_3_dist,model_3_price,model_3_eta,model_3_rank = np.zeros((n,)), np.zeros((n,)), np.zeros((n,)), np.zeros((n,))
model_4_dist,model_4_price,model_4_eta,model_4_rank = np.zeros((n,)), np.zeros((n,)), np.zeros((n,)), np.zeros((n,))
model_5_dist,model_5_price,model_5_eta,model_5_rank = np.zeros((n,)), np.zeros((n,)), np.zeros((n,)), np.zeros((n,))
model_6_dist,model_6_price,model_6_eta,model_6_rank = np.zeros((n,)), np.zeros((n,)), np.zeros((n,)), np.zeros((n,))
model_7_dist,model_7_price,model_7_eta,model_7_rank = np.zeros((n,)), np.zeros((n,)), np.zeros((n,)), np.zeros((n,))
model_8_dist,model_8_price,model_8_eta,model_8_rank = np.zeros((n,)), np.zeros((n,)), np.zeros((n,)), np.zeros((n,))
model_9_dist,model_9_price,model_9_eta,model_9_rank = np.zeros((n,)), np.zeros((n,)), np.zeros((n,)), np.zeros((n,))
model_10_dist,model_10_price,model_10_eta,model_10_rank = np.zeros((n,)), np.zeros((n,)), np.zeros((n,)), np.zeros((n,))
model_11_dist,model_11_price,model_11_eta,model_11_rank = np.zeros((n,)), np.zeros((n,)), np.zeros((n,)), np.zeros((n,))
n=0
for i, plan in tqdm(enumerate(data['plans_json'].values)):
if len(plan) == 0:
cur_plan_list = []
else:
cur_plan_list = plan
if len(cur_plan_list) == 0:
model_1_dist[i]=-1
model_1_price[i]=-1
model_1_eta[i]=-1
model_1_rank[i]=-1
model_2_dist[i]=-1
model_2_price[i]=-1
model_2_eta[i]=-1
model_2_rank[i]=-1
model_3_dist[i]=-1
model_3_price[i]=-1
model_3_eta[i]=-1
model_3_rank[i]=-1
model_4_dist[i]=-1
model_4_price[i]=-1
model_4_eta[i]=-1
model_4_rank[i]=-1
model_5_dist[i]=-1
model_5_price[i]=-1
model_5_eta[i]=-1
model_5_rank[i]=-1
model_6_dist[i]=-1
model_6_price[i]=-1
model_6_eta[i]=-1
model_6_rank[i]=-1
model_7_dist[i]=-1
model_7_price[i]=-1
model_7_eta[i]=-1
model_7_rank[i]=-1
model_8_dist[i]=-1
model_8_price[i]=-1
model_8_eta[i]=-1
model_8_rank[i]=-1
model_9_dist[i]=-1
model_9_price[i]=-1
model_9_eta[i]=-1
model_9_rank[i]=-1
model_10_dist[i]=-1
model_10_price[i]=-1
model_10_eta[i]=-1
model_10_rank[i]=-1
model_11_dist[i]=-1
model_11_price[i]=-1
model_11_eta[i]=-1
model_11_rank[i]=-1
else:
model_1_dist_list=[]
model_1_price_list=[]
model_1_eta_list=[]
model_1_rank_list=[]
model_2_dist_list=[]
model_2_price_list=[]
model_2_eta_list=[]
model_2_rank_list=[]
model_3_dist_list=[]
model_3_price_list=[]
model_3_eta_list=[]
model_3_rank_list=[]
model_4_dist_list=[]
model_4_price_list=[]
model_4_eta_list=[]
model_4_rank_list=[]
model_5_dist_list=[]
model_5_price_list=[]
model_5_eta_list=[]
model_5_rank_list=[]
model_6_dist_list=[]
model_6_price_list=[]
model_6_eta_list=[]
model_6_rank_list=[]
model_7_dist_list=[]
model_7_price_list=[]
model_7_eta_list=[]
model_7_rank_list=[]
model_8_dist_list=[]
model_8_price_list=[]
model_8_eta_list=[]
model_8_rank_list=[]
model_9_dist_list=[]
model_9_price_list=[]
model_9_eta_list=[]
model_9_rank_list=[]
model_10_dist_list=[]
model_10_price_list=[]
model_10_eta_list=[]
model_10_rank_list=[]
model_11_dist_list=[]
model_11_price_list=[]
model_11_eta_list=[]
model_11_rank_list=[]
mode_list = []
for tmp_dit in cur_plan_list:
if tmp_dit['price'] == '':
tmp_dit['price']=0
if tmp_dit['transport_mode']==1:
model_1_dist_list.append(int(tmp_dit['distance']))
model_1_price_list.append(int(tmp_dit['price']))
model_1_eta_list.append(int(tmp_dit['eta']))
elif tmp_dit['transport_mode']==2:
model_2_dist_list.append(int(tmp_dit['distance']))
model_2_price_list.append(int(tmp_dit['price']))
model_2_eta_list.append(int(tmp_dit['eta']))
elif tmp_dit['transport_mode']==3:
model_3_dist_list.append(int(tmp_dit['distance']))
model_3_price_list.append(int(tmp_dit['price']))
model_3_eta_list.append(int(tmp_dit['eta']))
elif tmp_dit['transport_mode']==4:
model_4_dist_list.append(int(tmp_dit['distance']))
model_4_price_list.append(int(tmp_dit['price']))
model_4_eta_list.append(int(tmp_dit['eta']))
elif tmp_dit['transport_mode']==5:
model_5_dist_list.append(int(tmp_dit['distance']))
model_5_price_list.append(int(tmp_dit['price']))
model_5_eta_list.append(int(tmp_dit['eta']))
elif tmp_dit['transport_mode']==6:
model_6_dist_list.append(int(tmp_dit['distance']))
model_6_price_list.append(int(tmp_dit['price']))
model_6_eta_list.append(int(tmp_dit['eta']))
elif tmp_dit['transport_mode']==7:
model_7_dist_list.append(int(tmp_dit['distance']))
model_7_price_list.append(int(tmp_dit['price']))
model_7_eta_list.append(int(tmp_dit['eta']))
elif tmp_dit['transport_mode']==8:
model_8_dist_list.append(int(tmp_dit['distance']))
model_8_price_list.append(int(tmp_dit['price']))
model_8_eta_list.append(int(tmp_dit['eta']))
elif tmp_dit['transport_mode']==9:
model_9_dist_list.append(int(tmp_dit['distance']))
model_9_price_list.append(int(tmp_dit['price']))
model_9_eta_list.append(int(tmp_dit['eta']))
elif tmp_dit['transport_mode']==10:
model_10_dist_list.append(int(tmp_dit['distance']))
model_10_price_list.append(int(tmp_dit['price']))
model_10_eta_list.append(int(tmp_dit['eta']))
elif tmp_dit['transport_mode']==11:
model_11_dist_list.append(int(tmp_dit['distance']))
model_11_price_list.append(int(tmp_dit['price']))
model_11_eta_list.append(int(tmp_dit['eta']))
mode_list.append(int(tmp_dit['transport_mode']))
mode_list.extend([1,2,3,4,5,6,7,8,9,10,11])
mode_list_end =len(mode_list)-11
model_1_dist[i]=np.mean(model_1_dist_list)
model_1_price[i]=np.mean(model_1_price_list)
model_1_eta[i]=np.mean(model_1_eta_list)
if mode_list.index(1)<mode_list_end:
model_1_rank[i]=mode_list.index(1)
else:
model_1_rank[i]=-1
model_2_dist[i]=np.mean(model_2_dist_list)
model_2_price[i]=np.mean(model_2_price_list)
model_2_eta[i]=np.mean(model_2_eta_list)
if mode_list.index(2)<mode_list_end:
model_2_rank[i]=mode_list.index(2)
else:
model_2_rank[i]=-1
model_3_dist[i]=np.mean(model_3_dist_list)
model_3_price[i]=np.mean(model_3_price_list)
model_3_eta[i]=np.mean(model_3_eta_list)
if mode_list.index(3)<mode_list_end:
model_3_rank[i]=mode_list.index(3)
else:
model_3_rank[i]=-1
model_4_dist[i]=np.mean(model_4_dist_list)
model_4_price[i]=np.mean(model_4_price_list)
model_4_eta[i]=np.mean(model_4_eta_list)
if mode_list.index(4)<mode_list_end:
model_4_rank[i]=mode_list.index(4)
else:
model_4_rank[i]=-1
model_5_dist[i]=np.mean(model_5_dist_list)
model_5_price[i]=np.mean(model_5_price_list)
model_5_eta[i]=np.mean(model_5_eta_list)
if mode_list.index(5)<mode_list_end:
model_5_rank[i]=mode_list.index(5)
else:
model_5_rank[i]=-1
model_6_dist[i]=np.mean(model_6_dist_list)
model_6_price[i]=np.mean(model_6_price_list)
model_6_eta[i]=np.mean(model_6_eta_list)
if mode_list.index(6)<mode_list_end:
model_6_rank[i]=mode_list.index(6)
else:
model_6_rank[i]=-1
model_7_dist[i]=np.mean(model_7_dist_list)
model_7_price[i]=np.mean(model_7_price_list)
model_7_eta[i]=np.mean(model_7_eta_list)
if mode_list.index(7)<mode_list_end:
model_7_rank[i]=mode_list.index(7)
else:
model_7_rank[i]=-1
model_8_dist[i]=np.mean(model_8_dist_list)
model_8_price[i]=np.mean(model_8_price_list)
model_8_eta[i]=np.mean(model_8_eta_list)
if mode_list.index(8)<mode_list_end:
model_8_rank[i]=mode_list.index(8)
else:
model_8_rank[i]=-1
model_9_dist[i]=np.mean(model_9_dist_list)
model_9_price[i]=np.mean(model_9_price_list)
model_9_eta[i]=np.mean(model_9_eta_list)
if mode_list.index(9)<mode_list_end:
model_9_rank[i]=mode_list.index(8)
else:
model_9_rank[i]=-1
model_10_dist[i]=np.mean(model_10_dist_list)
model_10_price[i]=np.mean(model_10_price_list)
model_10_eta[i]=np.mean(model_10_eta_list)
if mode_list.index(10)<mode_list_end:
model_10_rank[i]=mode_list.index(10)
else:
model_10_rank[i]=-1
model_11_dist[i]=np.mean(model_11_dist_list)
model_11_price[i]=np.mean(model_11_price_list)
model_11_eta[i]=np.mean(model_11_eta_list)
if mode_list.index(11)<mode_list_end:
model_11_rank[i]=mode_list.index(11)
else:
model_11_rank[i]=-1
feature_data = pd.DataFrame()
feature_data['plan_model_1_dist'] = model_1_dist
feature_data['plan_model_1_price'] = model_1_price
feature_data['plan_model_1_eta'] = model_1_eta
feature_data['plan_model_1_rank'] = model_1_rank
feature_data['plan_model_2_dist'] = model_2_dist
feature_data['plan_model_2_price'] = model_2_price
feature_data['plan_model_2_eta'] = model_2_eta
feature_data['plan_model_2_rank'] = model_2_rank
feature_data['plan_model_3_dist'] = model_3_dist
feature_data['plan_model_3_price'] = model_3_price
feature_data['plan_model_3_eta'] = model_3_eta
feature_data['plan_model_3_rank'] = model_3_rank
feature_data['plan_model_4_dist'] = model_4_dist
feature_data['plan_model_4_price'] = model_4_price
feature_data['plan_model_4_eta'] = model_4_eta
feature_data['plan_model_4_rank'] = model_4_rank
feature_data['plan_model_5_dist'] = model_5_dist
feature_data['plan_model_5_price'] = model_5_price
feature_data['plan_model_5_eta'] = model_5_eta
feature_data['plan_model_5_rank'] = model_5_rank
feature_data['plan_model_6_dist'] = model_6_dist
feature_data['plan_model_6_price'] = model_6_price
feature_data['plan_model_6_eta'] = model_6_eta
feature_data['plan_model_6_rank'] = model_6_rank
feature_data['plan_model_7_dist'] = model_7_dist
feature_data['plan_model_7_price'] = model_7_price
feature_data['plan_model_7_eta'] = model_7_eta
feature_data['plan_model_7_rank'] = model_7_rank
feature_data['plan_model_8_dist'] = model_8_dist
feature_data['plan_model_8_price'] = model_8_price
feature_data['plan_model_8_eta'] = model_8_eta
feature_data['plan_model_8_rank'] = model_8_rank
feature_data['plan_model_9_dist'] = model_9_dist
feature_data['plan_model_9_price'] = model_9_price
feature_data['plan_model_9_eta'] = model_9_eta
feature_data['plan_model_9_rank'] = model_9_rank
feature_data['plan_model_10_dist'] = model_10_dist
feature_data['plan_model_10_price'] = model_10_price
feature_data['plan_model_10_eta'] = model_10_eta
feature_data['plan_model_10_rank'] = model_10_rank
feature_data['plan_model_11_dist'] = model_11_dist
feature_data['plan_model_11_price'] = model_11_price
feature_data['plan_model_11_eta'] = model_11_eta
feature_data['plan_model_11_rank'] = model_11_rank
return feature_data
new_df=gen_plan_pingpu_feas(data)
data = pd.concat([data,new_df],axis=1)
pingpu_dist_feature=['plan_model_'+str(i)+'_dist' for i in range(1,12)]
pingpu_price_feature=['plan_model_'+str(i)+'_price' for i in range(1,12)]
pingpu_eta_feature=['plan_model_'+str(i)+'_eta' for i in range(1,12)]
pingpu_rank_feature=['plan_model_'+str(i)+'_rank' for i in range(1,12)]
plan_pingpu_feature = pingpu_dist_feature+pingpu_price_feature+pingpu_eta_feature+pingpu_rank_feature
print('plan推荐mode平铺特征数量：', len(plan_pingpu_feature), plan_pingpu_feature)
############################## 位置点和od对出现次数及排序特征 #################################
print('位置点和od对出现次数及排序特征')
#统计位置点出现的次数，并对次数进行大小的排序
#对地点出现次数进行排序，统计那些经常是导航起点的位置
o_appear_count=list(data.groupby(by=['o']).size())
d_appear_count=list(data.groupby(by=['d']).size())
the_query_o_count_df=pd.DataFrame()
the_query_d_count_df=pd.DataFrame()
o_list=[]
for name,group in data.groupby(by=['o']):
o_list.append(name)
d_list=[]
for name,group in data.groupby(by=['d']):
d_list.append(name)
the_query_o_count_df['o']=o_list
the_query_d_count_df['d']=d_list
the_query_o_count_df['o_appear_count']=o_appear_count
the_query_d_count_df['d_appear_count']=d_appear_count
data=data.merge(the_query_o_count_df, 'left', ['o'])
data=data.merge(the_query_d_count_df, 'left', ['d'])
data['o_appear_count_rank'] = data['o_appear_count'].rank() / float(data.shape[0])
data['d_appear_count_rank'] = data['d_appear_count'].rank() / float(data.shape[0])
data['o_appear_count_rank_buguiyi'] = data['o_appear_count'].rank()
data['d_appear_count_rank_buguiyi'] = data['d_appear_count'].rank()
#od共同出现情况及排序
#构建od对出现次数 统计特征，并对该特征进行排序，尝试效果。
data['od_couple']=data.apply(lambda row:(row['o']+'_'+row['d']),axis=1)
od_appear_count=list(data.groupby(by=['od_couple']).size())
the_query_od_count_df=pd.DataFrame()
od_list=[]
for name,group in data.groupby(by=['od_couple']):
od_list.append(name)
the_query_od_count_df['od_couple']=od_list
the_query_od_count_df['od_couple_count']=od_appear_count
data=data.merge(the_query_od_count_df, 'left', ['od_couple'])
data['od_couple_rank'] = data['od_couple_count'].rank() / float(data.shape[0])
data['od_couple_rank_buguiyi'] = data['od_couple_count'].rank()
od_appear_feature = ['o_appear_count','d_appear_count','o_appear_count_rank','d_appear_count_rank','o_appear_count_rank_buguiyi','d_appear_count_rank_buguiyi',
'od_couple','od_couple_count','od_couple_rank','od_couple_rank_buguiyi']
print('位置点和od对出现次数及排序特征数量：', len(od_appear_feature), od_appear_feature)
######################### 对plan_mode中的dis、eta、price和rank进行排序特征 ###########################
print('对plan_mode中的dis、eta、price和rank进行排序特征')
#第二波排序特征， 对平展后的每个特征进行整体的排序
for i in range(1,12):
data['plan_model_'+str(i)+'_dist'+'_rank']=data['plan_model_'+str(i)+'_dist'].rank()
data['plan_model_'+str(i)+'_eta'+'_rank']=data['plan_model_'+str(i)+'_eta'].rank()
data['plan_model_'+str(i)+'_price'+'_rank']=data['plan_model_'+str(i)+'_price'].rank()
data['plan_model_'+str(i)+'_rank'+'_rank']=data['plan_model_'+str(i)+'_rank'].rank()
data['plan_model_'+str(i)+'_dist'+'_rank_guiyi']=data['plan_model_'+str(i)+'_dist'].rank() / float(data.shape[0])
data['plan_model_'+str(i)+'_eta'+'_rank_guiyi']=data['plan_model_'+str(i)+'_eta'].rank() / float(data.shape[0])
data['plan_model_'+str(i)+'_price'+'_rank_guiyi']=data['plan_model_'+str(i)+'_price'].rank() / float(data.shape[0])
data['plan_model_'+str(i)+'_rank'+'_rank_guiyi']=data['plan_model_'+str(i)+'_rank'].rank() / float(data.shape[0])
pingzhan_dist_rank_feature=['plan_model_'+str(i)+'_dist_rank' for i in range(1,12)]
pingzhan_price_rank_feature=['plan_model_'+str(i)+'_price_rank' for i in range(1,12)]
pingzhan_eta_rank_feature=['plan_model_'+str(i)+'_eta_rank' for i in range(1,12)]
pingzhan_rank_rank_feature=['plan_model_'+str(i)+'_rank_rank' for i in range(1,12)]
pingzhan_dist_rank_guiyi_feature=['plan_model_'+str(i)+'_dist'+'_rank_guiyi' for i in range(1,12)]
pingzhan_price_rank_guiyi_feature=['plan_model_'+str(i)+'_eta'+'_rank_guiyi' for i in range(1,12)]
pingzhan_eta_rank_guiyi_feature=['plan_model_'+str(i)+'_price'+'_rank_guiyi' for i in range(1,12)]
pingzhan_rank_rank_guiyi_feature=['plan_model_'+str(i)+'_rank'+'_rank_guiyi' for i in range(1,12)]
plan_pingzhan_rank_feature=pingzhan_dist_rank_feature+pingzhan_price_rank_feature+pingzhan_eta_rank_feature+pingzhan_rank_rank_feature+pingzhan_dist_rank_guiyi_feature+pingzhan_price_rank_guiyi_feature+pingzhan_eta_rank_guiyi_feature+pingzhan_rank_rank_guiyi_feature
print('对plan_mode中的dis、eta、price和rank进行排序特征数量：', len(plan_pingzhan_rank_feature), plan_pingzhan_rank_feature)
########################################## 咕噜提分特征 #############################################
#试了，没提分，最后去掉了
print('咕噜提分特征')
def add_statistics(data, feat='pid'):
print('adding statistics ...')
# stat_data = data
# not contain val set
train_index = (data.req_time < '2018-11-23')
stat_data = data[train_index][['click_mode', feat]].reset_index(drop=True)
# static features
# each click mode for each feat, the number and the probability
feat_stat = stat_data[['click_mode', feat]]
feat_stat['feat_click_count'] = 1
# the number
feat_stat = feat_stat.groupby(['click_mode', feat]).agg('sum').reset_index()
feat_stat = feat_stat.sort_index(by=[feat, 'click_mode']).reset_index()
feat_sum = stat_data[[feat]]
feat_sum['feat_sum'] = 1
feat_sum = feat_sum.groupby([feat]).agg('sum').reset_index()
feat_sum = feat_sum.sort_index(by=[feat]).reset_index()
feat_mode_arr = np.zeros((len(feat_sum['feat_sum']), 12), dtype=np.float)
idx = 0
for r_i, row in tqdm(feat_stat.iterrows()):
# print(row)
if int(row[feat]) != feat_sum[feat].values[idx]:
idx += 1
# the probability
feat_mode_arr[idx, int(row['click_mode'])] = row['feat_click_count'] / feat_sum['feat_sum'].values[idx]
feat_mode = {}
feat_mode[feat] = feat_sum[feat]
for i in range(12):
feat_mode[feat + '_mode_' + str(i)] = feat_mode_arr[:, i]
feat_mode = pd.DataFrame(feat_mode)
data = data.merge(feat_mode, on=feat, how='left')
return data
data = add_statistics(data, feat='pid')
add_feature = ['pid' + '_mode_' + str(i) for i in range(1,12)]
print('咕噜提分特征数量：', len(add_feature), add_feature)
########################################## 协同特征 #############################################
xietong_data = pd.read_csv('../AdditionalData/xietong_feature_shanghai.csv', index_col = 0)
data = pd.concat([data,xietong_data],axis = 1)
print('特征加载')
###################################### 非特征 ######################################
or_feature = ['req_time','click_mode','sid']
###################################### 原始特征 ######################################
cate_feature = ['pid']
profile_feature = ['p' + str(i) for i in range(66)]
origin_num_feature = ['o_lng', 'o_lat', 'd_lng', 'd_lat'] + profile_feature + cate_feature
######################################## 经纬度特征 ##########################################
od_feature = ['od_manhattan_distance', 'od_manhattan_distance_detail', 'o_main_centroid_mean_dis', 'd_main_centroid_mean_dis', 'o_main_centroid_mode_dis', 'd_main_centroid_mode_dis', 'o_cluster', 'd_cluster']
######################################## 距离特征 ##########################################
distance_feature = ['manhattan','euclidean','delta_longitude','delta_latitude','pickup_x','pickup_y','pickup_z','dropoff_x','dropoff_y','dropoff_z','direction']
################################ plan_time & req_time特征 ###################################
#所有时间特征
all_time_feature = ['req_time_hour', 'req_time_weekday', 'req_time_minute', 'req_time_date_d', 'time_diff',
'diff_6_cloc', 'diff_12_clock', 'diff_18_clock', 'diff_24_clock']
#选择的时间特征
time_feature=['req_time_hour', 'time_diff', 'diff_6_cloc', 'diff_12_clock', 'diff_18_clock', 'diff_24_clock']
################################ 对profile的降维提取特征 ###################################
#所有特征
all_profiles_svd_feature = ['svd_fea_{}'.format(i) for i in range(10)]
#选择特征
profiles_svd_feature = []
############################## plans百度地图推荐交通方式特征 #################################
plan_feature = ['mode_feas_0', 'mode_feas_1', 'mode_feas_2', 'mode_feas_3', 'mode_feas_4', 'mode_feas_5', 'mode_feas_6', 'mode_feas_7', 'mode_feas_8', 'mode_feas_9', 'mode_feas_10', 'mode_feas_11',
'max_dist', 'min_dist', 'mean_dist', 'std_dist', 'max_price', 'min_price', 'mean_price', 'std_price', 'max_eta', 'min_eta', 'mean_eta', 'std_eta', 'max_dist_mode', 'min_dist_mode', 'max_price_mode', 'min_price_mode', 'max_eta_mode', 'min_eta_mode',
'first_mode', 'svd_mode_0', 'svd_mode_1', 'svd_mode_2', 'svd_mode_3', 'svd_mode_4', 'svd_mode_5', 'svd_mode_6', 'svd_mode_7', 'svd_mode_8', 'svd_mode_9']
############################## plan推荐mode平铺特征 #################################
pingpu_dist_feature=['plan_model_'+str(i)+'_dist' for i in range(1,12)]
pingpu_price_feature=['plan_model_'+str(i)+'_price' for i in range(1,12)]
pingpu_eta_feature=['plan_model_'+str(i)+'_eta' for i in range(1,12)]
pingpu_rank_feature=['plan_model_'+str(i)+'_rank' for i in range(1,12)]
#所有特征
all_plan_pingpu_feature = pingpu_dist_feature+pingpu_price_feature+pingpu_eta_feature+pingpu_rank_feature
#选择特征
plan_pingpu_feature = pingpu_dist_feature+pingpu_price_feature+pingpu_rank_feature
############################## 位置点和od对出现次数及排序特征 #################################
#所有特征
all_od_appear_feature = ['o_appear_count','d_appear_count','o_appear_count_rank','d_appear_count_rank','o_appear_count_rank_buguiyi','d_appear_count_rank_buguiyi',
'od_couple','od_couple_count','od_couple_rank','od_couple_rank_buguiyi']
#选择特征
od_appear_feature = ['o_appear_count','d_appear_count','o_appear_count_rank','d_appear_count_rank','o_appear_count_rank_buguiyi','d_appear_count_rank_buguiyi',
'od_couple_count']
######################### 对plan_mode中的dis、eta、price和rank进行排序特征 ###########################
pingzhan_dist_rank_feature=['plan_model_'+str(i)+'_dist_rank' for i in range(1,12)]
pingzhan_price_rank_feature=['plan_model_'+str(i)+'_price_rank' for i in range(1,12)]
pingzhan_eta_rank_feature=['plan_model_'+str(i)+'_eta_rank' for i in range(1,12)]
pingzhan_rank_rank_feature=['plan_model_'+str(i)+'_rank_rank' for i in range(1,12)]
pingzhan_dist_rank_guiyi_feature=['plan_model_'+str(i)+'_dist'+'_rank_guiyi' for i in range(1,12)]
pingzhan_price_rank_guiyi_feature=['plan_model_'+str(i)+'_eta'+'_rank_guiyi' for i in range(1,12)]
pingzhan_eta_rank_guiyi_feature=['plan_model_'+str(i)+'_price'+'_rank_guiyi' for i in range(1,12)]
pingzhan_rank_rank_guiyi_feature=['plan_model_'+str(i)+'_rank'+'_rank_guiyi' for i in range(1,12)]
#所有特征
all_plan_pingzhan_rank_feature=pingzhan_dist_rank_feature+pingzhan_price_rank_feature+pingzhan_eta_rank_feature+pingzhan_rank_rank_feature+pingzhan_dist_rank_guiyi_feature+pingzhan_price_rank_guiyi_feature+pingzhan_eta_rank_guiyi_feature+pingzhan_rank_rank_guiyi_feature
#选择特征
plan_pingzhan_rank_feature=pingzhan_dist_rank_feature+pingzhan_price_rank_feature+pingzhan_eta_rank_feature+pingzhan_rank_rank_feature
########################################## 咕噜提分特征 #############################################
add_feature = ['pid' + '_mode_' + str(i) for i in range(1,12)]
########################################## 协同特征 #############################################
xietong_feature = [col for col in xietong_data.columns]
########################################## 地铁特征 #############################################
########################################## 公交特征 #############################################
########################################## 大学特征 #############################################
########################################## 天气特征 #############################################
print('特征加载完成')
#保存选择的特征
Select_features = or_feature + origin_num_feature + od_feature + distance_feature + time_feature + profiles_svd_feature + plan_feature + all_plan_pingpu_feature + od_appear_feature + plan_pingzhan_rank_feature + add_feature + xietong_feature
#删除一部分 低重要度
Select_features.remove('plan_model_6_price_rank')
Select_features.remove('plan_model_5_price_rank')
Select_features.remove('plan_model_5_price')
Select_features.remove('plan_model_6_price')
Select_features.remove('plan_model_3_price_rank')
Select_features.remove('plan_model_3_price')
Select_features.remove('plan_model_9_rank_rank')
Select_features.remove('plan_model_6_rank_rank')
Select_features.remove('plan_model_5_rank_rank')
Select_features.remove('plan_model_8_rank_rank')
Select_features.remove('plan_model_10_rank_rank')
Select_features.remove('plan_model_11_rank_rank')
#使用的特征
feature = [col for col in Select_features if col not in ['req_time','click_mode','sid', 'pid']]
add_feature = ['pid' + '_mode_' + str(i) for i in range(1,12)]
feature =[col for col in feature if col not in add_feature ]
data[['p'+str(i) for i in range(66)]]= data[['p'+str(i) for i in range(66)]].fillna(-999)
######################################################################################
def f1_weighted_0(labels,preds):
preds = np.argmax(preds.reshape(12, -1), axis=0)
#print('labels:',labels)
#print('preds:',preds)
score = f1_score(y_true=labels, y_pred=preds, average='weighted')
return 'f1_weighted', score, True
def pred_proba_0(proba):
res = []
# pred_proba.tolist()
for i, e in enumerate(proba):
if e[3] >= 0.180: #0.2
e[3] = 1
if e[4] >= 0.118: #0.135
e[4] = 1
if e[6] >= 0.217:
e[6] = 1
if e[8] >= 0.21:
e[8] = 1
if e[9] >= 0.36:
e[9] = 1
if e[10] >= 0.285:
e[10]=1
if e[11] >= 0.34:
e[11]=1
res.append(e)
df = pd.DataFrame(res)
pred = df.idxmax(axis = 1)
# pred =
# return pred
return pred
###########################################训练#########################################
train_index = (data.req_time < '2018-11-23')
train_x = data[train_index][feature].reset_index(drop=True)
train_y = data[train_index].click_mode.reset_index(drop=True)
valid_index = (data.req_time > '2018-11-23') & (data.req_time < '2018-12-01')
valid_x = data[valid_index][feature].reset_index(drop=True)
valid_y = data[valid_index].click_mode.reset_index(drop=True)
test_index = (data.req_time > '2018-12-01')
test_x = data[test_index][feature].reset_index(drop=True)
print(len(feature), feature)
lgb_model_0 = lgb.LGBMClassifier(boosting_type="gbdt", num_leaves=100, reg_alpha=0, reg_lambda=0,
max_depth=-1, n_estimators=2000, objective='multiclass',
subsample=0.8, colsample_bytree=0.8, subsample_freq=1,min_child_samples = 50, learning_rate=0.05, random_state=35, metric="None",n_jobs=-1)
eval_set = [(valid_x, valid_y)]
lgb_model_0.fit(train_x, train_y, eval_set=eval_set, eval_metric=f1_weighted_0, verbose=10, early_stopping_rounds=200)
proba_all = lgb_model_0.predict_proba(data[feature])
proba_all = pd.DataFrame(proba_all)
prob_feature = [col for col in proba_all.columns]
data= pd.concat([data,proba_all],axis=1)
################################# 将多分类概率值加入二分类中 ##################################################
feature = feature + prob_feature
print('3 拆分plans构建二分类数据============================================')
data['plans_json'] = data['plans'].fillna('[]').apply(lambda x: json.loads(x))
plans = np.dstack((np.repeat(data['sid'].values,list(map(len,data['plans_json'].values))),np.concatenate(data['plans_json'].values)))
plans = pd.DataFrame(plans[0],columns=['sid','plans'])
plans = pd.concat([plans, pd.DataFrame.from_dict(plans['plans'].tolist())], axis=1)
plans.drop_duplicates(subset = ['sid','transport_mode'], inplace=True)
# plans['price'] = plans[['price','transport_mode']].apply(fillna_price, axis=1)
print('3.1 构建二分类plans的特征=============================================')
plans['price'] = plans['price'].apply(lambda x:0 if x == '' else int(x))
plans['price_yuan'] = plans['price']//100
plans['distance_eta'] = plans['distance']/plans['eta']
plans['distance_price'] = plans['distance']*plans['price']
plans['eta_price'] = plans['eta'] * plans['price']
plans['self_rank'] = plans.groupby(['sid']).cumcount()
plans['price_rank'] = plans.groupby(['sid'])['price'].rank()
plans['eta_rank'] = plans.groupby(['sid'])['eta'].rank()
plans['distance_rank'] = plans.groupby(['sid'])['distance'].rank()
plans['distance_eta_rank'] = plans.groupby(['sid'])['distance_eta'].rank()
plans['distance_price_price'] = plans.groupby(['sid'])['distance_price'].rank()