-
Notifications
You must be signed in to change notification settings - Fork 0
/
datatransform.py
1675 lines (1417 loc) · 178 KB
/
datatransform.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
'''
Transform Data
'''
# Import packages
import numpy as np
import pandas as pd
import os.path
import sys
import argparse
#import datasource as ds # to connect to BIM
from pyspark.context import SparkContext
from pyspark.sql import HiveContext, SparkSession
#find the year migration was made
def cohort_df(df, cpc_list):
'''
returns the cpc cohort each row belongs to. cohort 0 = belongs to no cohort
input: dataframe
output: dataframe or 1(failed=no cohorts found)
effects: modifies input df to add new cohort column
'''
#iterate through all cpc columns and update cohort
#Assumes the cpc list comes in ascending order of year. Eg. (cpc201209, cpc201210, etc.)
for cpc_year in range(1,len(cpc_list)):
if cpc_year ==1:
df['cohort'] = np.where(df[cpc_list[cpc_year-1]] != df[cpc_list[cpc_year]], int(cpc_list[cpc_year].strip('cpc')), 0)
else:
df['cohort'] = np.where(df[cpc_list[cpc_year-1]] != df[cpc_list[cpc_year]], int(cpc_list[cpc_year].strip('cpc')), df.cohort)
#to save space (int 64 to int 32)
df.cohort = df.cohort.astype('int32')
#No Cohorts found
if not df['cohort'].any():
print('No Cohort Found')
return df
def production_flag_cohort(df):
'''
Adds tag which identifies accounts to be used for production purposes (Target customers to run the model on) (Non Training or Testing)
Conditions for prod_tag:
1) No Migrations in the past year
2) Has a credit score > 680
3) Does NOT currently have TAW
4) Is NOT pending approval for a CC
5) Is NOT delinquent > 30 Days
'''
from datetime import date
import datetime
#Uncomment the commented lines below when the data collection is up to date. Currently the most recent cpc date is cpc201908
#df['cpc_before']=np.where(df['match_cpc_after'].isnull(), df['cpc' + (datetime.datetime.now().strftime('%Y')+datetime.datetime.now().strftime('%m'))], df.cpc_before)
#Comment the line below when data collection is up to date
df['cpc_before']=np.where(df['match_cpc_after'].isnull(), df['cpc201908'], df.cpc_before)
#today_year = np.uint32(date.today().year)
#today_month = np.uint32(date.today().month)
#df['cohort_month'] = np.where(df['cohort'] == 0, today_month, df.cohort_month)
#df['cohort_year'] = np.where(df['cohort'] == 0, today_year, df.cohort_year)
#df['cohort'] = np.where(df['cohort'] == 0, np.uint32(datetime.datetime.now().strftime('%Y')+datetime.datetime.now().strftime('%m')), df.cohort)
#df['cohort']=np.where(df['cohort'] == 0, np.uint32(201908), df.cohort)
#df['cohort_month'] = np.where(df['cohort'] == 0, np.uint32(8), df.cohort_month)
#df['cohort_year'] = np.where(df['cohort'] == 0, np.uint32(2019), df.cohort_year)
#check if most recent credit score is low
#uncomment bottom when data collection is set up to date
#df['low_credit_score'] = np.where(df['cr_bureau_score'+str(datetime.datetime.now().strftime('%Y')+datetime.datetime.now().strftime('%m'))] < 680, 1, 0)
df['low_credit_score'] = np.where(df['cr_bureau_score201908'] < 680, 1, 0)
#check if most recent cpc is TAW
#uncomment bottom when data collection is set up to date
#df['non_TAW'] = np.where(df['cpc'+str(datetime.datetime.now().strftime('%Y')+datetime.datetime.now().strftime('%m'))] =='TAW', 1, 0)
df['non_taw'] = np.where(df['cpc201908'] == 'TAW', 0, 1)
#Add production data tag
#df['match_prod_tag'] = np.where((df['non_TAW']==1) & (df['low_credit_score']==0) & (df['no_mig']==1)
# & (df['delinquent']==1) & (df['pending']==1), 1, 0)
#Add production data tag
df['match_prod_tag'] = np.where((df['non_taw']==1) & (df['low_credit_score']==0) & (df['no_mig']==1)
, 1, 0)
#remove unnecessary columns
df = df.drop(['non_taw','low_credit_score'], axis=1)
return df
'''
#Add delinquent tag , delinquent days > 30
with ds.connect('bim') as con:
sql = """
SELECT tsys_acct_id
into #TMP2
FROM [BIM_DAILY].[dbo].<TABLE_name>
"""
#Execute the SQL query (notice that code below is indented to the right as it is written under the 'with' statement)
con.execute(sql)
#Read data from the temporary table to a dataframe
df_delinquent = con.read_table('#TMP2')
df['delinquent'] = np.where(df['tsys_acct_id'].isin(df_delinquent['tsys_acct_id']), 1, 0)
#Identify the accounts pending approval for new CC
with ds.connect('bim') as con:
sql = """
SELECT tsys_acct_id
into #TMP1
FROM [BIM_DAILY].[dbo].[ACQUISITION_DAILY]
where App_Status = 'PD'
and tsys_acct_id is not NULL
and tsys_acct_id != 0
"""
#Execute the SQL query (notice that code below is indented to the right as it is written under the 'with' statement)
con.execute(sql)
#Read data from the temporary table to a dataframe
df_pending = con.read_table('#TMP1')
df['pending'] = np.where(df['tsys_acct_id'].isin(df_pending['tsys_acct_id']), 1, 0)
'''
# Define Migration Types of the CC. Eg. TPT-TIC, etc.
def Migration_type(df, cpc_list):
'''
returns the migration each row belongs to. migration '' = belongs to no migration
input: dataframe
output: dataframe or 1(failed=no migrations found)
effects: modifies input df to add new migrations column
'''
global mig_typ_found
mig_typ_found = True
#iterate through all cpc columns and update migration type
for cpc_year in range(1,len(cpc_list)):
if cpc_year ==1:
df['mig_typ'] = np.where(df[cpc_list[cpc_year-1]] != df[cpc_list[cpc_year]], df[cpc_list[cpc_year-1]]+'-'+df[cpc_list[cpc_year]], '')
else:
df['mig_typ'] = np.where(df[cpc_list[cpc_year-1]] != df[cpc_list[cpc_year]], df[cpc_list[cpc_year-1]]+'-'+df[cpc_list[cpc_year]], df.mig_typ)
#No migrations found
if not df['mig_typ'].any():
print('No Migration Types Found')
mig_typ_found = False
return df
def cpc_before_after(df, cpc_list):
'''
Split the Mig_typ column as cpc_before and cpc_After, and replace MMC to TPB. Drop Mig_typ column
'''
if mig_typ_found:
df[['cpc_before','match_cpc_after']] = df.mig_typ.str.split("-", n=3, expand=True).replace('MMC' , 'TPB')
else:
df['cpc_before'] = df[cpc_list[-1]]
df['match_cpc_after'] = None
df = df.drop('mig_typ',axis=1)
return df
def no_migration_flag(df, cpc_list):
'''
Adds columns to input dataframe where no cpc migration was made .Example, TIC-TIC. ie, # of unique cpcs per customer = 1
Input: DataFrame
Output: Mutated Dataframe
Effect: Mutates input dataframe
'''
df['no_mig'] = np.where(num_unique_per_row(df[cpc_list])==1, 1, 0)
return df
def remove_no_migration(df):
'''
Removes rows from input dataframe where no cpc migration was made & Data is not for production. Example, TIC-TIC
Input: DataFrame
Output: Mutated Dataframe
Effect: Mutates input dataframe
'''
df = df[(df['no_mig']==0)|(df['match_prod_tag']==1)]
df = df.drop('no_mig',axis=1)
return df
def num_unique_per_row(a):
'''
Efficient way to identify the number of unique elements per row in the input dataframe.
100X more efficient than pandas nunique() for large number of rows.
Helper function for remove_pure_gamers
Input: dataframe
Output: Frequency of unique element pandas dataframe
'''
b = np.sort(a,axis=1)
return pd.DataFrame((b[:,1:] != b[:,:-1]).sum(axis=1)+1)
def remove_pure_gamers(df,cpc_list):
'''
Removing customers who move from product A to B to C in the past year. ie, have >2 Unique products in a year
Input: DataFrame columns should only be cpc_list
'''
df['pure_gamers'] = np.where(num_unique_per_row(df[cpc_list])>2, 1, 0)
#drop pure gamers and the newly created tag
df = df[df.pure_gamers==0].drop('pure_gamers', axis=1)
return df
def tag_softgamers(df, cpc_list):
'''
Add Tag for customers who move from product A to B back to A in the past year.
Input: DataFrame,cpc_list
Note: call this function after adding no_mig tag
'''
df['soft_gamers'] = np.where((df[cpc_list[0]]==df[cpc_list[-1]]) & (df.no_mig==0), 1, 0)
return df
def remove_aeroplan_after(df):
'''
Remove rows from input dataframe where the customer is moving into an Aeroplan card
Input: Dataframe
Output: Dataframe
Effect: mutate input dataframe
'''
df = df[(df.match_cpc_after!='TAW')|(df.match_cpc_after!='TAC')|(df.match_cpc_after!='TAI')]
return df
def remove_emerald(df):
'''
Remove all rows from input dataframe where customer is moving from or to an Emerald Card
Input: Dataframe
Output: Dataframe
Effect: mutates input dataframe
'''
df = df[(df.cpc_before != 'TEF') | (df.cpc_before != 'TEV') | (df.match_cpc_after != 'TEF') | (df.match_cpc_after != 'TEV')]
return df
def remove_old_cards(df):
'''
Remove all rows from input dataframe where customer moved from or to an OLD card. (TDR & TCT)
'''
df = df[(df.cpc_before != 'TDR') | (df.cpc_before != 'TCT') | (df.match_cpc_after != 'TDR') | (df.match_cpc_after != 'TCT')]
return df
def dummy_cpc_before(df):
'''
One Hot Encode (Dummify) cpc_before and Drop cpc_before column from inout dataframe
'''
df = pd.concat([df, pd.get_dummies(df.cpc_before, prefix='match_cpc_before')], axis=1).drop('cpc_before', axis=1)
return df
def opendate_to_days(df):
'''
Adds days since Account Open Date to the input dataframe
input:dataframe
output: dataframe
effect: mutates dataframe
'''
from datetime import date
today = date.today()
# Adjusted the format
Open_Dt = pd.to_datetime(df.acct_open_dt,format ='%d%b%Y:%H:%M:%S')
# Setting Today with the same date format
Today = pd.to_datetime(today,format='%Y/%m/%d')
# Calculate the days passed
Dys_Pass = Today-Open_Dt
# Merge the pass day with the dataset
# to_numeric messes up the days:: changed to fix that and also store as int32 memory efficient
df.loc[:,'match_days_open'] = Dys_Pass.dt.days.astype('int32')
#Drop unneeded columns
df = df.drop('acct_open_dt', axis=1)
return df
def acct_index_pre_move(df):
'''
mutate input dataframe to add cols which match chq_ind, loc_ind, employee_ind, sav_ind, tfsa_ind, usd_chq_ind to the cohort
input: dataframe
output: dataframe or 1 (no indexes matched)
effect: mutates input dataframe
'''
#init lists with all chq_ind, loc_ind, employee_ind, sav_ind, tfsa_ind, usd_chq_ind columns from df
chq_list = list(df.filter(regex='chq_ind').columns)
loc_list = list(df.filter(regex='loc_ind').columns)
emp_list = list(df.filter(regex='employee_ind').columns)
sav_list = list(df.filter(regex='sav_ind').columns)
tfsa_list = list(df.filter(regex='tfsa_ind').columns)
usd_chq_list = list(df.filter(regex='usd_chq_ind').columns)
#init list of all unique cohorts
cohort_list = list(np.sort(df.cohort.unique()))
#iterate through all possible cohorts and create new cols for match index given the cohort matches
for cohort_ind in range(len(cohort_list)):
if cohort_ind == 0:
df['match_chq_ind'] = np.where(df.cohort == cohort_list[cohort_ind], df['chq_ind'+str(cohort_list[cohort_ind])], 0)
df['match_loc_ind'] = np.where(df.cohort == cohort_list[cohort_ind], df['loc_ind'+str(cohort_list[cohort_ind])], 0)
df['match_employee_ind'] = np.where(df.cohort == cohort_list[cohort_ind], df['employee_ind'+str(cohort_list[cohort_ind])], 0)
df['match_sav_ind'] = np.where(df.cohort == cohort_list[cohort_ind], df['sav_ind'+str(cohort_list[cohort_ind])], 0)
df['match_tfsa_ind'] = np.where(df.cohort == cohort_list[cohort_ind], df['tfsa_ind'+str(cohort_list[cohort_ind])], 0)
df['match_usd_chq_ind'] = np.where(df.cohort == cohort_list[cohort_ind], df['usd_chq_ind'+str(cohort_list[cohort_ind])], 0)
else:
df['match_chq_ind'] = np.where(df.cohort == cohort_list[cohort_ind], df['chq_ind'+str(cohort_list[cohort_ind])], df.match_chq_ind)
df['match_loc_ind'] = np.where(df.cohort == cohort_list[cohort_ind], df['loc_ind'+str(cohort_list[cohort_ind])], df.match_loc_ind)
df['match_employee_ind'] = np.where(df.cohort == cohort_list[cohort_ind], df['employee_ind'+str(cohort_list[cohort_ind])], df.match_employee_ind)
df['match_sav_ind'] = np.where(df.cohort == cohort_list[cohort_ind], df['sav_ind'+str(cohort_list[cohort_ind])], df.match_sav_ind)
df['match_tfsa_ind'] = np.where(df.cohort == cohort_list[cohort_ind], df['tfsa_ind'+str(cohort_list[cohort_ind])], df.match_tfsa_ind)
df['match_usd_chq_ind'] = np.where(df.cohort == cohort_list[cohort_ind], df['usd_chq_ind'+str(cohort_list[cohort_ind])], df.match_usd_chq_ind)
#set all the new columns to int32 instead of int64 to double efficiency
df['match_chq_ind'] = df['match_chq_ind'].fillna(0).astype('int32')
df['match_loc_ind'] = df['match_loc_ind'].fillna(0).astype('int32')
df['match_employee_ind'] = df['match_employee_ind'].fillna(0).astype('int32')
df['match_sav_ind'] = df['match_sav_ind'].fillna(0).astype('int32')
df['match_tfsa_ind'] = df['match_tfsa_ind'].fillna(0).astype('int32')
df['match_usd_chq_ind'] = df['match_usd_chq_ind'].fillna(0).astype('int32')
#Remove index columns not required anymore
df = df.drop(chq_list+loc_list+emp_list+sav_list+tfsa_list+usd_chq_list, axis = 1)
return df
def match_Amt(df):
'''
mutate input dataframe to add cols which match last annual fee amount, last autopayment amt, last change date, last credit limit change to the cohort
input: dataframe
output: dataframe or 1 (no last amounts matched)
effect: mutates input dataframe
'''
#init lists with all last annual fee amount, last autopayment amt, last change date, last credit limit change columns from df
annfee_list = list(df.filter(regex='last_ann_fee_amt').columns)
autopay_list = list(df.filter(regex='last_autopayment_amt').columns)
changedt_list = list(df.filter(regex='last_change_dt').columns)
credlim_list = list(df.filter(regex='lcrlim_change').columns)
#init list of all unique cohorts
cohort_list = list(np.sort(df.cohort.unique()))
#iterate through all possible cohorts and create new cols for match index given the cohort matches
for cohort_ind in range(len(cohort_list)):
if cohort_ind == 0:
df['match_last_ann_fee_amt'] = np.where(df.cohort == cohort_list[cohort_ind], df['last_ann_fee_amt'+str(cohort_list[cohort_ind])], 0)
df['match_last_autopayment_amt'] = np.where(df.cohort == cohort_list[cohort_ind], df['last_autopayment_amt'+str(cohort_list[cohort_ind])], 0)
df['match_last_change_dt'] = np.where(df.cohort == cohort_list[cohort_ind], df['last_change_dt'+str(cohort_list[cohort_ind])], 0)
df['match_lcrlim_change'] = np.where(df.cohort == cohort_list[cohort_ind], df['lcrlim_change'+str(cohort_list[cohort_ind])], 0)
else:
df['match_last_ann_fee_amt'] = np.where(df.cohort == cohort_list[cohort_ind], df['last_ann_fee_amt'+str(cohort_list[cohort_ind])], df.match_last_ann_fee_amt)
df['match_last_autopayment_amt'] = np.where(df.cohort == cohort_list[cohort_ind], df['last_autopayment_amt'+str(cohort_list[cohort_ind])], df.match_last_autopayment_amt)
df['match_last_change_dt'] = np.where(df.cohort == cohort_list[cohort_ind], df['last_change_dt'+str(cohort_list[cohort_ind])], df.match_last_change_dt)
df['match_lcrlim_change'] = np.where(df.cohort == cohort_list[cohort_ind], df['lcrlim_change'+str(cohort_list[cohort_ind])], df.match_lcrlim_change)
#set all the new columns to int32 instead of int64 to double efficiency
df['match_last_ann_fee_amt'] = df['match_last_ann_fee_amt'].fillna(0).astype('int32')
df['match_last_autopayment_amt'] = df['match_last_autopayment_amt'].astype('int32')
df['match_last_change_dt'] = df['match_last_change_dt'].astype('int32')
df['match_lcrlim_change'] = np.where(df['match_lcrlim_change']==0, df['acct_open_dt'], df['match_lcrlim_change'])
# Transform date cred limit change to days passed since the day
from datetime import date
today = date.today()
# Setting Today with the same date format
Today = pd.to_datetime(today,format='%Y/%m/%d')
lcrlim_change = pd.to_datetime(df.match_lcrlim_change,format ='%d%b%Y:%H:%M:%S')
# Calculate the pass day
lcrlim_dys_pass = Today - lcrlim_change
# to_numeric messes up the days:: changed to fix that and also store as int32 memory efficient
df.loc[:,'match_lcrlim_dys_pass'] = lcrlim_dys_pass.dt.days.astype('int32')
# Transform change date to days passed since the day
Change_dt = pd.to_datetime(df.match_last_change_dt,format ='%Y%M')
# Calculate the pass day
Change_Dys_Pass = Today - Change_dt
# to_numeric messes up the days:: changed to fix that and also store as int32 memory efficient
df.loc[:,'match_change_date_dys_pass'] = Change_Dys_Pass.dt.days.astype('int32')
#drop unneeded columns
df = df.drop(annfee_list+autopay_list+changedt_list+credlim_list+['match_lcrlim_change', 'match_last_change_dt'], axis = 1)
return df
def avg_three_months_prior_financial(df):
'''
Adds features like adb, Net Revenue, nibt, Amt Pymt Applied, Amt Revolved, Credit Limit Amt to the input dataframe. Avg of last 3 months
prior to migration
input: dataframe
output: dataframe
effect: mutates input dataframe
'''
#init lists with all adb, netrev , nibt, Amt Paid, Amt Revolved, Cred limit columns from df
adb_list = list(df.filter(regex='adb').columns)
netrev_list = list(df.filter(regex='net_revenue').columns)
nibt_list = list(df.filter(regex='nibt').columns)
amtpaid_list = list(df.filter(regex='amt_pymt_applied').columns)
amtrevolve_list = list(df.filter(regex='amt_revolve').columns)
credlim_list = list(df.filter(regex='credit_limit_amt').columns)
#init list of all unique cohorts
cohort_list = list(np.sort(df.cohort.unique()))
#iterate through all possible cohorts and create new cols for match index given the cohort matches
for cohort_ind in range(len(cohort_list)):
#check if cohort in january or february
january = '01' in str(cohort_list[cohort_ind])[-2:]
feb = '02' in str(cohort_list[cohort_ind])[-2:]
march = '03' in str(cohort_list[cohort_ind])[-2:]
year = str(cohort_list[cohort_ind])[:-2]
if january:
if cohort_ind == 0:
df['match_adb_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+str(int(year)-1)+'10']+df['adb'+str(int(year)-1)+'12']+df['adb'+str(int(year)-1)+'11'])/3, 0)
df['match_netrev_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['net_revenue'+str(int(year)-1)+'10']+ df['net_revenue'+str(int(year)-1)+'12']+ df['net_revenue'+str(int(year)-1)+'11'])/3, 0)
df['match_nibt_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nibt'+str(int(year)-1)+'10']+df['nibt'+str(int(year)-1)+'12']+df['nibt'+str(int(year)-1)+'11'])/3, 0)
df['match_amtpaid_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_pymt_applied'+str(int(year)-1)+'10']+df['amt_pymt_applied'+str(int(year)-1)+'12']+df['amt_pymt_applied'+str(int(year)-1)+'11'])/3, 0)
df['match_amtrevolve_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_revolve'+str(int(year)-1)+'10']+df['amt_revolve'+str(int(year)-1)+'12']+df['amt_revolve'+str(int(year)-1)+'11'])/3, 0)
df['match_credlimit_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['credit_limit_amt'+str(int(year)-1)+'10']+df['credit_limit_amt'+str(int(year)-1)+'12']+df['credit_limit_amt'+str(int(year)-1)+'11'])/3, 0)
df['match_adb_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+str(int(year)-1)+'10']-df['adb'+str(int(year)-1)+'11']), 0)
df['match_netrev_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['net_revenue'+str(int(year)-1)+'10']-df['net_revenue'+str(int(year)-1)+'11']), 0)
df['match_nibt_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nibt'+str(int(year)-1)+'10']-df['nibt'+str(int(year)-1)+'11']), 0)
df['match_amtpaid_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_pymt_applied'+str(int(year)-1)+'10']-df['amt_pymt_applied'+str(int(year)-1)+'11']), 0)
df['match_amtrevolve_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_revolve'+str(int(year)-1)+'10']-df['amt_revolve'+str(int(year)-1)+'11']), 0)
df['match_credlimit_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['credit_limit_amt'+str(int(year)-1)+'10']-df['credit_limit_amt'+str(int(year)-1)+'11']), 0)
df['match_adb_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+str(int(year)-1)+'11']-df['adb'+str(int(year)-1)+'12']), 0)
df['match_netrev_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['net_revenue'+str(int(year)-1)+'11']-df['net_revenue'+str(int(year)-1)+'12']), 0)
df['match_nibt_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nibt'+str(int(year)-1)+'11']-df['nibt'+str(int(year)-1)+'12']), 0)
df['match_amtpaid_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_pymt_applied'+str(int(year)-1)+'11']-df['amt_pymt_applied'+str(int(year)-1)+'12']), 0)
df['match_amtrevolve_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_revolve'+str(int(year)-1)+'11']-df['amt_revolve'+str(int(year)-1)+'12']), 0)
df['match_credlimit_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['credit_limit_amt'+str(int(year)-1)+'11']-df['credit_limit_amt'+str(int(year)-1)+'12']), 0)
else:
df['match_adb_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+str(int(year)-1)+'10']+df['adb'+str(int(year)-1)+'12']+df['adb'+str(int(year)-1)+'11'])/3, df.match_adb_prev_3months)
df['match_netrev_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['net_revenue'+str(int(year)-1)+'10']+ df['net_revenue'+str(int(year)-1)+'12']+ df['net_revenue'+str(int(year)-1)+'11'])/3, df.match_netrev_prev_3months)
df['match_nibt_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nibt'+str(int(year)-1)+'10']+df['nibt'+str(int(year)-1)+'12']+df['nibt'+str(int(year)-1)+'11'])/3, df.match_nibt_prev_3months)
df['match_amtpaid_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_pymt_applied'+str(int(year)-1)+'10']+df['amt_pymt_applied'+str(int(year)-1)+'12']+df['amt_pymt_applied'+str(int(year)-1)+'11'])/3, df.match_amtpaid_prev_3months)
df['match_amtrevolve_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_revolve'+str(int(year)-1)+'10']+df['amt_revolve'+str(int(year)-1)+'12']+df['amt_revolve'+str(int(year)-1)+'11'])/3, df.match_amtrevolve_prev_3months)
df['match_credlimit_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['credit_limit_amt'+str(int(year)-1)+'10']+df['credit_limit_amt'+str(int(year)-1)+'12']+df['credit_limit_amt'+str(int(year)-1)+'11'])/3, df.match_credlimit_prev_3months)
df['match_adb_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+str(int(year)-1)+'10']-df['adb'+str(int(year)-1)+'11']), df.match_adb_prev_3months_delta_3_2)
df['match_netrev_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['net_revenue'+str(int(year)-1)+'10']-df['net_revenue'+str(int(year)-1)+'11']), df.match_netrev_prev_3months_delta_3_2)
df['match_nibt_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nibt'+str(int(year)-1)+'10']-df['nibt'+str(int(year)-1)+'11']), df.match_nibt_prev_3months_delta_3_2)
df['match_amtpaid_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_pymt_applied'+str(int(year)-1)+'10']-df['amt_pymt_applied'+str(int(year)-1)+'11']), df.match_amtpaid_prev_3months_delta_3_2)
df['match_amtrevolve_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_revolve'+str(int(year)-1)+'10']-df['amt_revolve'+str(int(year)-1)+'11']), df.match_amtrevolve_prev_3months_delta_3_2)
df['match_credlimit_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['credit_limit_amt'+str(int(year)-1)+'10']-df['credit_limit_amt'+str(int(year)-1)+'11']), df.match_credlimit_prev_3months_delta_3_2)
df['match_adb_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+str(int(year)-1)+'11']-df['adb'+str(int(year)-1)+'12']), df.match_adb_prev_3months_delta_2_1)
df['match_netrev_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['net_revenue'+str(int(year)-1)+'11']-df['net_revenue'+str(int(year)-1)+'12']), df.match_netrev_prev_3months_delta_2_1)
df['match_nibt_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nibt'+str(int(year)-1)+'11']-df['nibt'+str(int(year)-1)+'12']), df.match_nibt_prev_3months_delta_2_1)
df['match_amtpaid_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_pymt_applied'+str(int(year)-1)+'11']-df['amt_pymt_applied'+str(int(year)-1)+'12']), df.match_amtpaid_prev_3months_delta_2_1)
df['match_amtrevolve_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_revolve'+str(int(year)-1)+'11']-df['amt_revolve'+str(int(year)-1)+'12']), df.match_amtrevolve_prev_3months_delta_2_1)
df['match_credlimit_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['credit_limit_amt'+str(int(year)-1)+'11']-df['credit_limit_amt'+str(int(year)-1)+'12']), df.match_credlimit_prev_3months_delta_2_1)
elif feb:
if cohort_ind == 0:
df['match_adb_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+str(int(year)-1)+'11']+df['adb'+year+'01']+df['adb'+str(int(year)-1)+'12'])/3, 0)
df['match_netrev_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['net_revenue'+str(int(year)-1)+'11']+ df['net_revenue'+year+'01']+ df['net_revenue'+str(int(year)-1)+'12'])/3, 0)
df['match_nibt_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nibt'+str(int(year)-1)+'11']+df['nibt'+year+'01']+df['nibt'+str(int(year)-1)+'12'])/3, 0)
df['match_amtpaid_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_pymt_applied'+str(int(year)-1)+'11']+df['amt_pymt_applied'+year+'01']+df['amt_pymt_applied'+str(int(year)-1)+'12'])/3, 0)
df['match_amtrevolve_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_revolve'+str(int(year)-1)+'11']+df['amt_revolve'+year+'01']+df['amt_revolve'+str(int(year)-1)+'12'])/3, 0)
df['match_credlimit_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['credit_limit_amt'+str(int(year)-1)+'11']+df['credit_limit_amt'+year+'01']+df['credit_limit_amt'+str(int(year)-1)+'12'])/3, 0)
df['match_adb_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+str(int(year)-1)+'11']-df['adb'+str(int(year)-1)+'12']), 0)
df['match_netrev_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['net_revenue'+str(int(year)-1)+'11']-df['net_revenue'+str(int(year)-1)+'12']), 0)
df['match_nibt_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nibt'+str(int(year)-1)+'11']-df['nibt'+str(int(year)-1)+'12']), 0)
df['match_amtpaid_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_pymt_applied'+str(int(year)-1)+'11']-df['amt_pymt_applied'+str(int(year)-1)+'12']), 0)
df['match_amtrevolve_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_revolve'+str(int(year)-1)+'11']-df['amt_revolve'+str(int(year)-1)+'12']), 0)
df['match_credlimit_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['credit_limit_amt'+str(int(year)-1)+'11']-df['credit_limit_amt'+str(int(year)-1)+'12']), 0)
df['match_adb_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+str(int(year)-1)+'12']-df['adb'+year+'01']), 0)
df['match_netrev_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['net_revenue'+str(int(year)-1)+'12']-df['net_revenue'+year+'01']), 0)
df['match_nibt_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nibt'+str(int(year)-1)+'12']-df['nibt'+year+'01']), 0)
df['match_amtpaid_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_pymt_applied'+str(int(year)-1)+'12']-df['amt_pymt_applied'+year+'01']), 0)
df['match_amtrevolve_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_revolve'+str(int(year)-1)+'12']-df['amt_revolve'+year+'01']), 0)
df['match_credlimit_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['credit_limit_amt'+str(int(year)-1)+'12']-df['credit_limit_amt'+year+'01']), 0)
else:
df['match_adb_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+str(int(year)-1)+'11']+df['adb'+year+'01']+df['adb'+str(int(year)-1)+'12'])/3, df.match_adb_prev_3months)
df['match_netrev_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['net_revenue'+str(int(year)-1)+'11']+ df['net_revenue'+year+'01']+ df['net_revenue'+str(int(year)-1)+'12'])/3, df.match_netrev_prev_3months)
df['match_nibt_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nibt'+str(int(year)-1)+'11']+df['nibt'+year+'01']+df['nibt'+str(int(year)-1)+'12'])/3, df.match_nibt_prev_3months)
df['match_amtpaid_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_pymt_applied'+str(int(year)-1)+'11']+df['amt_pymt_applied'+year+'01']+df['amt_pymt_applied'+str(int(year)-1)+'12'])/3, df.match_amtpaid_prev_3months)
df['match_amtrevolve_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_revolve'+str(int(year)-1)+'11']+df['amt_revolve'+year+'01']+df['amt_revolve'+str(int(year)-1)+'12'])/3, df.match_amtrevolve_prev_3months)
df['match_credlimit_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['credit_limit_amt'+str(int(year)-1)+'11']+df['credit_limit_amt'+year+'01']+df['credit_limit_amt'+str(int(year)-1)+'12'])/3, df.match_credlimit_prev_3months)
df['match_adb_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+str(int(year)-1)+'11']-df['adb'+str(int(year)-1)+'12']), df.match_adb_prev_3months_delta_3_2)
df['match_netrev_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['net_revenue'+str(int(year)-1)+'11']-df['net_revenue'+str(int(year)-1)+'12']), df.match_netrev_prev_3months_delta_3_2)
df['match_nibt_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nibt'+str(int(year)-1)+'11']-df['nibt'+str(int(year)-1)+'12']), df.match_nibt_prev_3months_delta_3_2)
df['match_amtpaid_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_pymt_applied'+str(int(year)-1)+'11']-df['amt_pymt_applied'+str(int(year)-1)+'12']), df.match_amtpaid_prev_3months_delta_3_2)
df['match_amtrevolve_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_revolve'+str(int(year)-1)+'11']-df['amt_revolve'+str(int(year)-1)+'12']), df.match_amtrevolve_prev_3months_delta_3_2)
df['match_credlimit_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['credit_limit_amt'+str(int(year)-1)+'11']-df['credit_limit_amt'+str(int(year)-1)+'12']), df.match_credlimit_prev_3months_delta_3_2)
df['match_adb_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+str(int(year)-1)+'12']-df['adb'+year+'01']), df.match_adb_prev_3months_delta_2_1)
df['match_netrev_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['net_revenue'+str(int(year)-1)+'12']-df['net_revenue'+year+'01']), df.match_netrev_prev_3months_delta_2_1)
df['match_nibt_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nibt'+str(int(year)-1)+'12']-df['nibt'+year+'01']), df.match_nibt_prev_3months_delta_2_1)
df['match_amtpaid_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_pymt_applied'+str(int(year)-1)+'12']-df['amt_pymt_applied'+year+'01']), df.match_amtpaid_prev_3months_delta_2_1)
df['match_amtrevolve_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_revolve'+str(int(year)-1)+'12']-df['amt_revolve'+year+'01']), df.match_amtrevolve_prev_3months_delta_2_1)
df['match_credlimit_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['credit_limit_amt'+str(int(year)-1)+'12']-df['credit_limit_amt'+year+'01']), df.match_credlimit_prev_3months_delta_2_1)
elif march:
if cohort_ind == 0:
df['match_adb_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+year+'02']+df['adb'+year+'01']+df['adb'+str(int(year)-1)+'12'])/3, 0)
df['match_netrev_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['net_revenue'+year+'02']+ df['net_revenue'+year+'01']+ df['net_revenue'+str(int(year)-1)+'12'])/3, 0)
df['match_nibt_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nibt'+year+'02']+df['nibt'+year+'01']+df['nibt'+str(int(year)-1)+'12'])/3, 0)
df['match_amtpaid_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_pymt_applied'+year+'02']+df['amt_pymt_applied'+year+'01']+df['amt_pymt_applied'+str(int(year)-1)+'12'])/3, 0)
df['match_amtrevolve_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_revolve'+year+'02']+df['amt_revolve'+year+'01']+df['amt_revolve'+str(int(year)-1)+'12'])/3, 0)
df['match_credlimit_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['credit_limit_amt'+year+'02']+df['credit_limit_amt'+year+'01']+df['credit_limit_amt'+str(int(year)-1)+'12'])/3, 0)
df['match_adb_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+str(int(year)-1)+'12']-df['adb'+year+'01']), 0)
df['match_netrev_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['net_revenue'+str(int(year)-1)+'12']-df['net_revenue'+year+'01']), 0)
df['match_nibt_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nibt'+str(int(year)-1)+'12']-df['nibt'+year+'01']), 0)
df['match_amtpaid_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_pymt_applied'+str(int(year)-1)+'12']-df['amt_pymt_applied'+year+'01']), 0)
df['match_amtrevolve_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_revolve'+str(int(year)-1)+'12']-df['amt_revolve'+year+'01']), 0)
df['match_credlimit_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['credit_limit_amt'+str(int(year)-1)+'12']-df['credit_limit_amt'+year+'01']), 0)
df['match_adb_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+year+'01']-df['adb'+year+'02']), 0)
df['match_netrev_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+year+'01']-df['net_revenue'+year+'02']), 0)
df['match_nibt_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+year+'01']-df['nibt'+year+'02']), 0)
df['match_amtpaid_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+year+'01']-df['amt_pymt_applied'+year+'02']), 0)
df['match_amtrevolve_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+year+'01']-df['amt_revolve'+year+'02']), 0)
df['match_credlimit_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+year+'01']-df['credit_limit_amt'+year+'02']), 0)
else:
df['match_adb_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+year+'02']+df['adb'+year+'01']+df['adb'+str(int(year)-1)+'12'])/3, df.match_adb_prev_3months)
df['match_netrev_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['net_revenue'+year+'02']+ df['net_revenue'+year+'01']+ df['net_revenue'+str(int(year)-1)+'12'])/3, df.match_netrev_prev_3months)
df['match_nibt_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nibt'+year+'02']+df['nibt'+year+'01']+df['nibt'+str(int(year)-1)+'12'])/3, df.match_nibt_prev_3months)
df['match_amtpaid_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_pymt_applied'+year+'02']+df['amt_pymt_applied'+year+'01']+df['amt_pymt_applied'+str(int(year)-1)+'12'])/3, df.match_amtpaid_prev_3months)
df['match_amtrevolve_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_revolve'+year+'02']+df['amt_revolve'+year+'01']+df['amt_revolve'+str(int(year)-1)+'12'])/3, df.match_amtrevolve_prev_3months)
df['match_credlimit_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['credit_limit_amt'+year+'02']+df['credit_limit_amt'+year+'01']+df['credit_limit_amt'+str(int(year)-1)+'12'])/3, df.match_credlimit_prev_3months)
df['match_adb_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+str(int(year)-1)+'12']-df['adb'+year+'01']), df.match_adb_prev_3months_delta_3_2)
df['match_netrev_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['net_revenue'+str(int(year)-1)+'12']-df['net_revenue'+year+'01']), df.match_netrev_prev_3months_delta_3_2)
df['match_nibt_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nibt'+str(int(year)-1)+'12']-df['nibt'+year+'01']), df.match_nibt_prev_3months_delta_3_2)
df['match_amtpaid_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_pymt_applied'+str(int(year)-1)+'12']-df['amt_pymt_applied'+year+'01']), df.match_amtpaid_prev_3months_delta_3_2)
df['match_amtrevolve_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_revolve'+str(int(year)-1)+'12']-df['amt_revolve'+year+'01']), df.match_amtrevolve_prev_3months_delta_3_2)
df['match_credlimit_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['credit_limit_amt'+str(int(year)-1)+'12']-df['credit_limit_amt'+year+'01']), df.match_credlimit_prev_3months_delta_3_2)
df['match_adb_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+year+'01']-df['adb'+year+'02']), df.match_adb_prev_3months_delta_2_1)
df['match_netrev_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+year+'01']-df['net_revenue'+year+'02']), df.match_netrev_prev_3months_delta_2_1)
df['match_nibt_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+year+'01']-df['nibt'+year+'02']), df.match_nibt_prev_3months_delta_2_1)
df['match_amtpaid_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+year+'01']-df['amt_pymt_applied'+year+'02']), df.match_amtpaid_prev_3months_delta_2_1)
df['match_amtrevolve_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+year+'01']-df['amt_revolve'+year+'02']), df.match_amtrevolve_prev_3months_delta_2_1)
df['match_credlimit_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+year+'01']-df['credit_limit_amt'+year+'02']), df.match_credlimit_prev_3months_delta_2_1)
else:
if cohort_ind == 0:
df['match_adb_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+str(cohort_list[cohort_ind]-3)]+df['adb'+str(cohort_list[cohort_ind]-1)]+df['adb'+str(cohort_list[cohort_ind]-2)])/3, 0)
df['match_netrev_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['net_revenue'+str(cohort_list[cohort_ind]-3)]+ df['net_revenue'+str(cohort_list[cohort_ind]-1)]+ df['net_revenue'+str(cohort_list[cohort_ind]-2)])/3, 0)
df['match_nibt_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nibt'+str(cohort_list[cohort_ind]-3)]+df['nibt'+str(cohort_list[cohort_ind]-1)]+df['nibt'+str(cohort_list[cohort_ind]-2)])/3, 0)
df['match_amtpaid_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_pymt_applied'+str(cohort_list[cohort_ind]-3)]+df['amt_pymt_applied'+str(cohort_list[cohort_ind]-1)]+df['amt_pymt_applied'+str(cohort_list[cohort_ind]-2)])/3, 0)
df['match_amtrevolve_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_revolve'+str(cohort_list[cohort_ind]-3)]+df['amt_revolve'+str(cohort_list[cohort_ind]-1)]+df['amt_revolve'+str(cohort_list[cohort_ind]-2)])/3, 0)
df['match_credlimit_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['credit_limit_amt'+str(cohort_list[cohort_ind]-3)]+df['credit_limit_amt'+str(cohort_list[cohort_ind]-1)]+df['credit_limit_amt'+str(cohort_list[cohort_ind]-2)])/3, 0)
df['match_adb_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+str(cohort_list[cohort_ind]-3)]-df['adb'+str(cohort_list[cohort_ind]-2)]), 0)
df['match_netrev_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['net_revenue'+str(cohort_list[cohort_ind]-3)]-df['net_revenue'+str(cohort_list[cohort_ind]-2)]), 0)
df['match_nibt_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nibt'+str(cohort_list[cohort_ind]-3)]-df['nibt'+str(cohort_list[cohort_ind]-2)]), 0)
df['match_amtpaid_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_pymt_applied'+str(cohort_list[cohort_ind]-3)]-df['amt_pymt_applied'+str(cohort_list[cohort_ind]-2)]), 0)
df['match_amtrevolve_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_revolve'+str(cohort_list[cohort_ind]-3)]-df['amt_revolve'+str(cohort_list[cohort_ind]-2)]), 0)
df['match_credlimit_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['credit_limit_amt'+str(cohort_list[cohort_ind]-3)]-df['credit_limit_amt'+str(cohort_list[cohort_ind]-2)]), 0)
df['match_adb_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+str(cohort_list[cohort_ind]-2)]-df['adb'+str(cohort_list[cohort_ind]-1)]), 0)
df['match_netrev_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['net_revenue'+str(cohort_list[cohort_ind]-2)]-df['net_revenue'+str(cohort_list[cohort_ind]-1)]), 0)
df['match_nibt_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nibt'+str(cohort_list[cohort_ind]-2)]-df['nibt'+str(cohort_list[cohort_ind]-1)]), 0)
df['match_amtpaid_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_pymt_applied'+str(cohort_list[cohort_ind]-2)]-df['amt_pymt_applied'+str(cohort_list[cohort_ind]-1)]), 0)
df['match_amtrevolve_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_revolve'+str(cohort_list[cohort_ind]-2)]-df['amt_revolve'+str(cohort_list[cohort_ind]-1)]), 0)
df['match_credlimit_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['credit_limit_amt'+str(cohort_list[cohort_ind]-2)]-df['credit_limit_amt'+str(cohort_list[cohort_ind]-1)]), 0)
else:
df['match_adb_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+str(cohort_list[cohort_ind]-3)]+df['adb'+str(cohort_list[cohort_ind]-1)]+df['adb'+str(cohort_list[cohort_ind]-2)])/3, df.match_adb_prev_3months)
df['match_netrev_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['net_revenue'+str(cohort_list[cohort_ind]-3)]+df['net_revenue'+str(cohort_list[cohort_ind]-1)]+df['net_revenue'+str(cohort_list[cohort_ind]-2)])/3, df.match_netrev_prev_3months)
df['match_nibt_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nibt'+str(cohort_list[cohort_ind]-3)]+df['nibt'+str(cohort_list[cohort_ind]-1)]+df['nibt'+str(cohort_list[cohort_ind]-2)])/3, df.match_nibt_prev_3months)
df['match_amtpaid_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_pymt_applied'+str(cohort_list[cohort_ind]-3)]+df['amt_pymt_applied'+str(cohort_list[cohort_ind]-1)]+df['amt_pymt_applied'+str(cohort_list[cohort_ind]-2)])/3, df.match_amtpaid_prev_3months)
df['match_amtrevolve_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_revolve'+str(cohort_list[cohort_ind]-3)]+df['amt_revolve'+str(cohort_list[cohort_ind]-1)]+df['amt_revolve'+str(cohort_list[cohort_ind]-2)])/3, df.match_amtrevolve_prev_3months)
df['match_credlimit_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['credit_limit_amt'+str(cohort_list[cohort_ind]-3)]+df['credit_limit_amt'+str(cohort_list[cohort_ind]-1)]+df['credit_limit_amt'+str(cohort_list[cohort_ind]-2)])/3, df.match_credlimit_prev_3months)
df['match_adb_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+str(cohort_list[cohort_ind]-3)]-df['adb'+str(cohort_list[cohort_ind]-2)]), df.match_adb_prev_3months_delta_3_2)
df['match_netrev_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['net_revenue'+str(cohort_list[cohort_ind]-3)]-df['net_revenue'+str(cohort_list[cohort_ind]-2)]), df.match_netrev_prev_3months_delta_3_2)
df['match_nibt_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nibt'+str(cohort_list[cohort_ind]-3)]-df['nibt'+str(cohort_list[cohort_ind]-2)]), df.match_nibt_prev_3months_delta_3_2)
df['match_amtpaid_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_pymt_applied'+str(cohort_list[cohort_ind]-3)]-df['amt_pymt_applied'+str(cohort_list[cohort_ind]-2)]), df.match_amtpaid_prev_3months_delta_3_2)
df['match_amtrevolve_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_revolve'+str(cohort_list[cohort_ind]-3)]-df['amt_revolve'+str(cohort_list[cohort_ind]-2)]), df.match_amtrevolve_prev_3months_delta_3_2)
df['match_credlimit_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['credit_limit_amt'+str(cohort_list[cohort_ind]-3)]-df['credit_limit_amt'+str(cohort_list[cohort_ind]-2)]), df.match_credlimit_prev_3months_delta_3_2)
df['match_adb_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['adb'+str(cohort_list[cohort_ind]-2)]-df['adb'+str(cohort_list[cohort_ind]-1)]), df.match_adb_prev_3months_delta_2_1)
df['match_netrev_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['net_revenue'+str(cohort_list[cohort_ind]-2)]-df['net_revenue'+str(cohort_list[cohort_ind]-1)]), df.match_netrev_prev_3months_delta_2_1)
df['match_nibt_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nibt'+str(cohort_list[cohort_ind]-2)]-df['nibt'+str(cohort_list[cohort_ind]-1)]), df.match_nibt_prev_3months_delta_2_1)
df['match_amtpaid_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_pymt_applied'+str(cohort_list[cohort_ind]-2)]-df['amt_pymt_applied'+str(cohort_list[cohort_ind]-1)]), df.match_amtpaid_prev_3months_delta_2_1)
df['match_amtrevolve_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['amt_revolve'+str(cohort_list[cohort_ind]-2)]-df['amt_revolve'+str(cohort_list[cohort_ind]-1)]), df.match_amtrevolve_prev_3months_delta_2_1)
df['match_credlimit_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['credit_limit_amt'+str(cohort_list[cohort_ind]-2)]-df['credit_limit_amt'+str(cohort_list[cohort_ind]-1)]), df.match_credlimit_prev_3months_delta_2_1)
#set all the new columns to int32 instead of int64
df['match_adb_prev_3months'] = df['match_adb_prev_3months'].astype('int32')
df['match_netrev_prev_3months'] = df['match_netrev_prev_3months'].astype('int32')
df['match_nibt_prev_3months'] = df['match_nibt_prev_3months'].astype('int32')
df['match_amtpaid_prev_3months'] = df['match_amtpaid_prev_3months'].astype('int32')
df['match_amtrevolve_prev_3months'] = df['match_amtrevolve_prev_3months'].astype('int32')
df['match_credlimit_prev_3months'] = df['match_credlimit_prev_3months'].astype('int32')
df['match_adb_prev_3months_delta_3_2'] = df['match_adb_prev_3months_delta_3_2'].astype('float32')
df['match_netrev_prev_3months_delta_3_2'] = df['match_netrev_prev_3months_delta_3_2'].astype('float32')
df['match_nibt_prev_3months_delta_3_2'] = df['match_nibt_prev_3months_delta_3_2'].astype('float32')
df['match_amtpaid_prev_3months_delta_3_2'] = df['match_amtpaid_prev_3months_delta_3_2'].astype('float32')
df['match_amtrevolve_prev_3months_delta_3_2'] = df['match_amtrevolve_prev_3months_delta_3_2'].astype('float32')
df['match_credlimit_prev_3months_delta_3_2'] = df['match_credlimit_prev_3months_delta_3_2'].astype('float32')
df['match_adb_prev_3months_delta_2_1'] = df['match_adb_prev_3months_delta_2_1'].astype('float32')
df['match_netrev_prev_3months_delta_2_1'] = df['match_netrev_prev_3months_delta_2_1'].astype('float32')
df['match_nibt_prev_3months_delta_2_1'] = df['match_nibt_prev_3months_delta_2_1'].astype('float32')
df['match_amtpaid_prev_3months_delta_2_1'] = df['match_amtpaid_prev_3months_delta_2_1'].astype('float32')
df['match_amtrevolve_prev_3months_delta_2_1'] = df['match_amtrevolve_prev_3months_delta_2_1'].astype('float32')
df['match_credlimit_prev_3months_delta_2_1'] = df['match_credlimit_prev_3months_delta_2_1'].astype('float32')
return df
###
# Need to do this for quarters and previous years later
###
def index_customer_basics(df):
'''
mutate input dataframe to add cols which match province_code, easyweb_ind, student_ind, secure_acct_ind to the cohort
input: dataframe
output: dataframe or 1 (no indexes matched)
effect: mutates input dataframe
'''
#init lists with all chq_ind, loc_ind, employee_ind, sav_ind, usd_chq_ind columns from df
#prov_list = list(df.filter(regex='Province_Cd').columns)
easyweb_list = list(df.filter(regex='easyweb_ind').columns)
student_list = list(df.filter(regex='student_ind').columns)
secureacct_list = list(df.filter(regex='secure_acct_ind').columns)
#init list of all unique cohorts
cohort_list = list(np.sort(df.cohort.unique()))
#iterate through all possible cohorts and create new cols for match index given the cohort matches
for cohort_ind in range(len(cohort_list)):
if cohort_ind == 0:
#df['match_Province_Cd'] = np.where(df.cohort == cohort_list[cohort_ind], df['Province_Cd'+str(cohort_list[cohort_ind])], 0)
df['match_easyweb_ind'] = np.where(df.cohort == cohort_list[cohort_ind], df['easyweb_ind'+str(cohort_list[cohort_ind])], 0)
df['match_student_ind'] = np.where(df.cohort == cohort_list[cohort_ind], df['student_ind'+str(cohort_list[cohort_ind])], 0)
df['match_secure_acct_ind'] = np.where(df.cohort == cohort_list[cohort_ind], df['secure_acct_ind'+str(cohort_list[cohort_ind])], 0)
else:
#df['match_Province_Cd'] = np.where(df.cohort == cohort_list[cohort_ind], df['Province_Cd'+str(cohort_list[cohort_ind])], df.match_Province_Cd)
df['match_easyweb_ind'] = np.where(df.cohort == cohort_list[cohort_ind], df['easyweb_ind'+str(cohort_list[cohort_ind])], df.match_easyweb_ind)
df['match_student_ind'] = np.where(df.cohort == cohort_list[cohort_ind], df['student_ind'+str(cohort_list[cohort_ind])], df.match_student_ind)
df['match_secure_acct_ind'] = np.where(df.cohort == cohort_list[cohort_ind], df['secure_acct_ind'+str(cohort_list[cohort_ind])], df.match_secure_acct_ind)
#set all the new columns to int32 instead of int64 to double efficiency
#df['match_Province_Cd'] = df['match_Province_Cd'].astype('int32')
df['match_easyweb_ind'] = df['match_easyweb_ind'].fillna(0).astype('int32')
df['match_student_ind'] = df['match_student_ind'].fillna(0).astype('int32')
df['match_secure_acct_ind'] = df['match_secure_acct_ind'].fillna(0).astype('int32')
#Remove index columns not required anymore
df = df.drop(easyweb_list+student_list+secureacct_list, axis = 1)
return df
def match_Ind(df):
'''
mutate input dataframe to add cols which match EGScore, Credit Score, FIW Flag, SOW Balance, SOW Cl, Triad Align Score, Inactive Months to the cohort
input: dataframe
output: dataframe or 1 (no indexes matched)
effect: mutates input dataframe
'''
EG_list = list(df.filter(regex='eg_score').columns)
CScore_list = list(df.filter(regex='cr_bureau_score').columns)
FIW_list = list(df.filter(regex='fiw_flg_').columns)
SOWBal_list = list(df.filter(regex='sow_bal_').columns)
SOWCl_list = list(df.filter(regex='sow_cl_').columns)
triad_align_score_list = list(df.filter(regex='triad_align_score').columns)
inactive_months_list = list(df.filter(regex='inactive_months').columns)
#init list of all unique cohorts
cohort_list = list(np.sort(df.cohort.unique()))
#iterate through all possible cohorts and create new cols for match index given the cohort matches
for cohort_ind in range(len(cohort_list)):
if cohort_ind == 0:
df['match_eg_score'] = np.where(df.cohort == cohort_list[cohort_ind], df['eg_score'+str(cohort_list[cohort_ind])], 0)
df['match_Credit_Score'] = np.where(df.cohort == cohort_list[cohort_ind], df['cr_bureau_score'+str(cohort_list[cohort_ind])], 0)
df['match_FIW_flg'] = np.where(df.cohort == cohort_list[cohort_ind], df['fiw_flg_'+str(cohort_list[cohort_ind])], 0)
df['match_SOW_Bal'] = np.where(df.cohort == cohort_list[cohort_ind], df['sow_bal_'+str(cohort_list[cohort_ind])], 0)
df['match_SOW_CL'] = np.where(df.cohort == cohort_list[cohort_ind], df['sow_cl_'+str(cohort_list[cohort_ind])], 0)
df['match_triad_align_score'] = np.where(df.cohort == cohort_list[cohort_ind], df['triad_align_score'+str(cohort_list[cohort_ind])], 0)
df['match_inactive_months'] = np.where(df.cohort == cohort_list[cohort_ind], df['inactive_months'+str(cohort_list[cohort_ind])], 0)
else:
df['match_eg_score'] = np.where(df.cohort == cohort_list[cohort_ind], df['eg_score'+str(cohort_list[cohort_ind])], df.match_eg_score)
df['match_Credit_Score'] = np.where(df.cohort == cohort_list[cohort_ind], df['cr_bureau_score'+str(cohort_list[cohort_ind])], df.match_Credit_Score)
df['match_FIW_flg'] = np.where(df.cohort == cohort_list[cohort_ind], df['fiw_flg_'+str(cohort_list[cohort_ind])], df.match_FIW_flg)
df['match_SOW_Bal'] = np.where(df.cohort == cohort_list[cohort_ind], df['sow_bal_'+str(cohort_list[cohort_ind])], df.match_SOW_Bal)
df['match_SOW_CL'] = np.where(df.cohort == cohort_list[cohort_ind], df['sow_cl_'+str(cohort_list[cohort_ind])], df.match_SOW_CL)
df['match_triad_align_score'] = np.where(df.cohort == cohort_list[cohort_ind], df['triad_align_score'+str(cohort_list[cohort_ind])], df.match_triad_align_score)
df['match_inactive_months'] = np.where(df.cohort == cohort_list[cohort_ind], df['inactive_months'+str(cohort_list[cohort_ind])], df.match_inactive_months)
#set all the new columns to int32 instead of int64 to double efficiency
df['match_eg_score'] = df['match_eg_score'].fillna(0).astype('int32')
df['match_Credit_Score'] = df['match_Credit_Score'].fillna(0).astype('int32')
df['match_FIW_flg'] = df['match_FIW_flg'].astype('int32')
df['match_SOW_Bal'] = df['match_SOW_Bal'].astype('int32')
df['match_SOW_CL'] = df['match_SOW_CL'].astype('int32')
df['match_triad_align_score'] = df['match_triad_align_score'].fillna(0).astype('int32')
df['match_inactive_months'] = df['match_inactive_months'].astype('int32')
#Remove index columns not required anymore
df = df.drop(EG_list+CScore_list+FIW_list+SOWBal_list+SOWCl_list+triad_align_score_list+inactive_months_list, axis = 1)
return df
def acct_cash_feats(df):
'''
Add Cash Adv Cnt, Cash Adv Amt, Cash Bal Amt, Bal Protection fee, NSF Fee, Overlimit Fee columns (Avg 3 months prior move) to input dataframe
input: Dataframe
output: Dataframe
mutates input dataframe
'''
#init lists with all Cash Adv Cnt, Cash Adv Amt, Cash Bal Amt, Bal Protection fee, NSF Fee, Overlimit Fee columns from df
cash_adv_count_list = list(df.filter(regex='cash_advance_count').columns)
cash_adv_bal_list = list(df.filter(regex='cash_advance_amt').columns)
Cash_Bal_list = list(df.filter(regex='cash_bal_amt').columns)
Bal_protect_list = list(df.filter(regex='bal_protection_fee').columns)
nsf_list = list(df.filter(regex='nsf_fee').columns)
olimit_fee_list = list(df.filter(regex='overlimit_fee').columns)
#init list of all unique cohorts
cohort_list = list(np.sort(df.cohort.unique()))
#iterate through all possible cohorts and create new cols for match index given the cohort matches
for cohort_ind in range(len(cohort_list)):
#check if cohort in january or february
january = '01' in str(cohort_list[cohort_ind])[-2:]
feb = '02' in str(cohort_list[cohort_ind])[-2:]
march = '03' in str(cohort_list[cohort_ind])[-2:]
year = str(cohort_list[cohort_ind])[:-2]
if january:
if cohort_ind == 0:
df['match_cash_adv_count_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['cash_advance_count'+str(int(year)-1)+'10']+df['cash_advance_count'+str(int(year)-1)+'12']+df['cash_advance_count'+str(int(year)-1)+'11'])/3, 0)
df['match_cash_adv_bal_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['cash_advance_amt'+str(int(year)-1)+'10']+ df['cash_advance_amt'+str(int(year)-1)+'12']+ df['cash_advance_amt'+str(int(year)-1)+'11'])/3, 0)
df['match_bal_protection_fee_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['cash_bal_amt'+str(int(year)-1)+'10']+df['cash_bal_amt'+str(int(year)-1)+'12']+df['cash_bal_amt'+str(int(year)-1)+'11'])/3, 0)
df['match_nsf_fee_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['bal_protection_fee'+str(int(year)-1)+'10']+df['bal_protection_fee'+str(int(year)-1)+'12']+df['bal_protection_fee'+str(int(year)-1)+'11'])/3, 0)
df['match_overlimit_fee_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nsf_fee'+str(int(year)-1)+'10']+df['nsf_fee'+str(int(year)-1)+'12']+df['nsf_fee'+str(int(year)-1)+'11'])/3, 0)
else:
df['match_cash_adv_count_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['cash_advance_count'+str(int(year)-1)+'10']+df['cash_advance_count'+str(int(year)-1)+'12']+df['cash_advance_count'+str(int(year)-1)+'11'])/3, df.match_cash_adv_count_prev_3months)
df['match_cash_adv_bal_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['cash_advance_amt'+str(int(year)-1)+'10']+ df['cash_advance_amt'+str(int(year)-1)+'12']+ df['cash_advance_amt'+str(int(year)-1)+'11'])/3, df.match_cash_adv_bal_prev_3months)
df['match_bal_protection_fee_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['cash_bal_amt'+str(int(year)-1)+'10']+df['cash_bal_amt'+str(int(year)-1)+'12']+df['cash_bal_amt'+str(int(year)-1)+'11'])/3, df.match_bal_protection_fee_prev_3months)
df['match_nsf_fee_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['bal_protection_fee'+str(int(year)-1)+'10']+df['bal_protection_fee'+str(int(year)-1)+'12']+df['bal_protection_fee'+str(int(year)-1)+'11'])/3, df.match_nsf_fee_prev_3months)
df['match_overlimit_fee_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nsf_fee'+str(int(year)-1)+'10']+df['nsf_fee'+str(int(year)-1)+'12']+df['nsf_fee'+str(int(year)-1)+'11'])/3, df.match_overlimit_fee_prev_3months)
elif feb:
if cohort_ind == 0:
df['match_cash_adv_count_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['cash_advance_count'+str(int(year)-1)+'11']+df['cash_advance_count'+year+'01']+df['cash_advance_count'+str(int(year)-1)+'12'])/3, 0)
df['match_cash_adv_bal_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['cash_advance_amt'+str(int(year)-1)+'11']+ df['cash_advance_amt'+year+'01']+ df['cash_advance_amt'+str(int(year)-1)+'12'])/3, 0)
df['match_bal_protection_fee_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['cash_bal_amt'+str(int(year)-1)+'11']+df['cash_bal_amt'+year+'01']+df['cash_bal_amt'+str(int(year)-1)+'12'])/3, 0)
df['match_nsf_fee_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['bal_protection_fee'+str(int(year)-1)+'11']+df['bal_protection_fee'+year+'01']+df['bal_protection_fee'+str(int(year)-1)+'12'])/3, 0)
df['match_overlimit_fee_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nsf_fee'+str(int(year)-1)+'11']+df['nsf_fee'+year+'01']+df['nsf_fee'+str(int(year)-1)+'12'])/3, 0)
else:
df['match_cash_adv_count_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['cash_advance_count'+str(int(year)-1)+'11']+df['cash_advance_count'+year+'01']+df['cash_advance_count'+str(int(year)-1)+'12'])/3, df.match_cash_adv_count_prev_3months)
df['match_cash_adv_bal_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['cash_advance_amt'+str(int(year)-1)+'11']+ df['cash_advance_amt'+year+'01']+ df['cash_advance_amt'+str(int(year)-1)+'12'])/3, df.match_cash_adv_bal_prev_3months)
df['match_bal_protection_fee_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['cash_bal_amt'+str(int(year)-1)+'11']+df['cash_bal_amt'+year+'01']+df['cash_bal_amt'+str(int(year)-1)+'12'])/3, df.match_bal_protection_fee_prev_3months)
df['match_nsf_fee_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['bal_protection_fee'+str(int(year)-1)+'11']+df['bal_protection_fee'+year+'01']+df['bal_protection_fee'+str(int(year)-1)+'12'])/3, df.match_nsf_fee_prev_3months)
df['match_overlimit_fee_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nsf_fee'+str(int(year)-1)+'11']+df['nsf_fee'+year+'01']+df['nsf_fee'+str(int(year)-1)+'12'])/3, df.match_overlimit_fee_prev_3months)
elif march:
if cohort_ind == 0:
df['match_cash_adv_count_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['cash_advance_count'+year+'02']+df['cash_advance_count'+year+'01']+df['cash_advance_count'+str(int(year)-1)+'12'])/3, 0)
df['match_cash_adv_bal_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['cash_advance_amt'+year+'02']+ df['cash_advance_amt'+year+'01']+ df['cash_advance_amt'+str(int(year)-1)+'12'])/3, 0)
df['match_bal_protection_fee_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['cash_bal_amt'+year+'02']+df['cash_bal_amt'+year+'01']+df['cash_bal_amt'+str(int(year)-1)+'12'])/3, 0)
df['match_nsf_fee_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['bal_protection_fee'+year+'02']+df['bal_protection_fee'+year+'01']+df['bal_protection_fee'+str(int(year)-1)+'12'])/3, 0)
df['match_overlimit_fee_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nsf_fee'+year+'02']+df['nsf_fee'+year+'01']+df['nsf_fee'+str(int(year)-1)+'12'])/3, 0)
else:
df['match_cash_adv_count_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['cash_advance_count'+year+'02']+df['cash_advance_count'+year+'01']+df['cash_advance_count'+str(int(year)-1)+'12'])/3, df.match_cash_adv_count_prev_3months)
df['match_cash_adv_bal_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['cash_advance_amt'+year+'02']+ df['cash_advance_amt'+year+'01']+ df['cash_advance_amt'+str(int(year)-1)+'12'])/3, df.match_cash_adv_bal_prev_3months)
df['match_bal_protection_fee_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['cash_bal_amt'+year+'02']+df['cash_bal_amt'+year+'01']+df['cash_bal_amt'+str(int(year)-1)+'12'])/3, df.match_bal_protection_fee_prev_3months)
df['match_nsf_fee_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['bal_protection_fee'+year+'02']+df['bal_protection_fee'+year+'01']+df['bal_protection_fee'+str(int(year)-1)+'12'])/3, df.match_nsf_fee_prev_3months)
df['match_overlimit_fee_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nsf_fee'+year+'02']+df['nsf_fee'+year+'01']+df['nsf_fee'+str(int(year)-1)+'12'])/3, df.match_overlimit_fee_prev_3months)
else:
if cohort_ind == 0:
df['match_cash_adv_count_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['cash_advance_count'+str(cohort_list[cohort_ind]-3)]+df['cash_advance_count'+str(cohort_list[cohort_ind]-1)]+df['cash_advance_count'+str(cohort_list[cohort_ind]-2)])/3, 0)
df['match_cash_adv_bal_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['cash_advance_amt'+str(cohort_list[cohort_ind]-3)]+ df['cash_advance_amt'+str(cohort_list[cohort_ind]-1)]+ df['cash_advance_amt'+str(cohort_list[cohort_ind]-2)])/3, 0)
df['match_bal_protection_fee_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['cash_bal_amt'+str(cohort_list[cohort_ind]-3)]+df['cash_bal_amt'+str(cohort_list[cohort_ind]-1)]+df['cash_bal_amt'+str(cohort_list[cohort_ind]-2)])/3, 0)
df['match_nsf_fee_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['bal_protection_fee'+str(cohort_list[cohort_ind]-3)]+df['bal_protection_fee'+str(cohort_list[cohort_ind]-1)]+df['bal_protection_fee'+str(cohort_list[cohort_ind]-2)])/3, 0)
df['match_overlimit_fee_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nsf_fee'+str(cohort_list[cohort_ind]-3)]+df['nsf_fee'+str(cohort_list[cohort_ind]-1)]+df['nsf_fee'+str(cohort_list[cohort_ind]-2)])/3, 0)
else:
df['match_cash_adv_count_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['cash_advance_count'+str(cohort_list[cohort_ind]-3)]+df['cash_advance_count'+str(cohort_list[cohort_ind]-1)]+df['cash_advance_count'+str(cohort_list[cohort_ind]-2)])/3, df.match_cash_adv_count_prev_3months)
df['match_cash_adv_bal_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['cash_advance_amt'+str(cohort_list[cohort_ind]-3)]+df['cash_advance_amt'+str(cohort_list[cohort_ind]-1)]+df['cash_advance_amt'+str(cohort_list[cohort_ind]-2)])/3, df.match_cash_adv_bal_prev_3months)
df['match_bal_protection_fee_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['cash_bal_amt'+str(cohort_list[cohort_ind]-3)]+df['cash_bal_amt'+str(cohort_list[cohort_ind]-1)]+df['cash_bal_amt'+str(cohort_list[cohort_ind]-2)])/3, df.match_bal_protection_fee_prev_3months)
df['match_nsf_fee_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['bal_protection_fee'+str(cohort_list[cohort_ind]-3)]+df['bal_protection_fee'+str(cohort_list[cohort_ind]-1)]+df['bal_protection_fee'+str(cohort_list[cohort_ind]-2)])/3, df.match_nsf_fee_prev_3months)
df['match_overlimit_fee_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['nsf_fee'+str(cohort_list[cohort_ind]-3)]+df['nsf_fee'+str(cohort_list[cohort_ind]-1)]+df['nsf_fee'+str(cohort_list[cohort_ind]-2)])/3, df.match_overlimit_fee_prev_3months)
#set all the new columns to int32 instead of int64
df['match_cash_adv_count_prev_3months'] = df['match_cash_adv_count_prev_3months'].astype('int32')
df['match_cash_adv_bal_prev_3months'] = df['match_cash_adv_bal_prev_3months'].astype('int32')
df['match_bal_protection_fee_prev_3months'] = df['match_bal_protection_fee_prev_3months'].astype('int32')
df['match_nsf_fee_prev_3months'] = df['match_nsf_fee_prev_3months'].astype('int32')
df['match_overlimit_fee_prev_3months'] = df['match_overlimit_fee_prev_3months'].astype('int32')
#Remove index columns not required anymore
df = df.drop(cash_adv_count_list+cash_adv_bal_list+Cash_Bal_list+Bal_protect_list+nsf_list+olimit_fee_list, axis = 1)
return df
def rewards_acct(df):
'''
Add Apple Pay Amount, Apple pay count , rewards accrued, aeroplan expense amt columns to input dataframe (Avg 3 months prior move)
input: dataframe
output: dataframe
effect: mutates input dataframe
'''
#init lists with all Apple Pay Amount, Apple pay count , rewards accrued, aeroplan expense amt columns from df
aero_amt_list = list(df.filter(regex='aeroplan_expense_amt').columns)
apple_pay_amt_list = list(df.filter(regex='apple_pay_purchase_amt').columns)
apple_pay_count_list = list(df.filter(regex='apple_pay_purchase_cnt').columns)
rewards_list = list(df.filter(regex='rewards_accrued').columns)
#init list of all unique cohorts
cohort_list = list(np.sort(df.cohort.unique()))
#iterate through all possible cohorts and create new cols for match index given the cohort matches
for cohort_ind in range(len(cohort_list)):
#check if cohort in january or february
january = '01' in str(cohort_list[cohort_ind])[-2:]
feb = '02' in str(cohort_list[cohort_ind])[-2:]
march = '03' in str(cohort_list[cohort_ind])[-2:]
year = str(cohort_list[cohort_ind])[:-2]
if january:
if cohort_ind == 0:
df['match_aeroplan_expense_amount_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['aeroplan_expense_amt'+str(int(year)-1)+'10']+df['aeroplan_expense_amt'+str(int(year)-1)+'12']+df['aeroplan_expense_amt'+str(int(year)-1)+'11'])/3, 0)
df['match_apple_pay_purchase_amount_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_amt'+str(int(year)-1)+'10']+ df['apple_pay_purchase_amt'+str(int(year)-1)+'12']+ df['apple_pay_purchase_amt'+str(int(year)-1)+'11'])/3, 0)
df['match_apple_pay_purchase_count_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_cnt'+str(int(year)-1)+'10']+df['apple_pay_purchase_cnt'+str(int(year)-1)+'12']+df['apple_pay_purchase_cnt'+str(int(year)-1)+'11'])/3, 0)
df['match_rewards_accrued_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['rewards_accrued'+str(int(year)-1)+'10']+df['rewards_accrued'+str(int(year)-1)+'12']+df['rewards_accrued'+str(int(year)-1)+'11'])/3, 0)
df['match_aeroplan_expense_amount_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['aeroplan_expense_amt'+str(int(year)-1)+'10']-df['aeroplan_expense_amt'+str(int(year)-1)+'11']), 0)
df['match_apple_pay_purchase_amount_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_amt'+str(int(year)-1)+'10']-df['apple_pay_purchase_amt'+str(int(year)-1)+'11']), 0)
df['match_apple_pay_purchase_count_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_cnt'+str(int(year)-1)+'10']-df['apple_pay_purchase_cnt'+str(int(year)-1)+'11']), 0)
df['match_rewards_accrued_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['rewards_accrued'+str(int(year)-1)+'10']-df['rewards_accrued'+str(int(year)-1)+'11']), 0)
df['match_aeroplan_expense_amount_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['aeroplan_expense_amt'+str(int(year)-1)+'11']-df['aeroplan_expense_amt'+str(int(year)-1)+'12']), 0)
df['match_apple_pay_purchase_amount_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_amt'+str(int(year)-1)+'11']-df['apple_pay_purchase_amt'+str(int(year)-1)+'12']), 0)
df['match_apple_pay_purchase_count_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_cnt'+str(int(year)-1)+'11']-df['apple_pay_purchase_cnt'+str(int(year)-1)+'12']), 0)
df['match_rewards_accrued_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['rewards_accrued'+str(int(year)-1)+'11']-df['rewards_accrued'+str(int(year)-1)+'12']), 0)
else:
df['match_aeroplan_expense_amount_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['aeroplan_expense_amt'+str(int(year)-1)+'10']+df['aeroplan_expense_amt'+str(int(year)-1)+'12']+df['aeroplan_expense_amt'+str(int(year)-1)+'11'])/3, df.match_aeroplan_expense_amount_prev_3months)
df['match_apple_pay_purchase_amount_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_amt'+str(int(year)-1)+'10']+ df['apple_pay_purchase_amt'+str(int(year)-1)+'12']+ df['apple_pay_purchase_amt'+str(int(year)-1)+'11'])/3, df.match_apple_pay_purchase_amount_prev_3months)
df['match_apple_pay_purchase_count_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_cnt'+str(int(year)-1)+'10']+df['apple_pay_purchase_cnt'+str(int(year)-1)+'12']+df['apple_pay_purchase_cnt'+str(int(year)-1)+'11'])/3, df.match_apple_pay_purchase_count_prev_3months)
df['match_rewards_accrued_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['rewards_accrued'+str(int(year)-1)+'10']+df['rewards_accrued'+str(int(year)-1)+'12']+df['rewards_accrued'+str(int(year)-1)+'11'])/3, df.match_rewards_accrued_3months)
df['match_aeroplan_expense_amount_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['aeroplan_expense_amt'+str(int(year)-1)+'10']-df['aeroplan_expense_amt'+str(int(year)-1)+'11']), df.match_aeroplan_expense_amount_prev_3months_delta_3_2)
df['match_apple_pay_purchase_amount_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_amt'+str(int(year)-1)+'10']-df['apple_pay_purchase_amt'+str(int(year)-1)+'11']), df.match_apple_pay_purchase_amount_prev_3months_delta_3_2)
df['match_apple_pay_purchase_count_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_cnt'+str(int(year)-1)+'10']-df['apple_pay_purchase_cnt'+str(int(year)-1)+'11']), df.match_apple_pay_purchase_count_prev_3months_delta_3_2)
df['match_rewards_accrued_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['rewards_accrued'+str(int(year)-1)+'10']-df['rewards_accrued'+str(int(year)-1)+'11']), df.match_rewards_accrued_3months_delta_3_2)
df['match_aeroplan_expense_amount_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['aeroplan_expense_amt'+str(int(year)-1)+'11']-df['aeroplan_expense_amt'+str(int(year)-1)+'12']), df.match_aeroplan_expense_amount_prev_3months_delta_2_1)
df['match_apple_pay_purchase_amount_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_amt'+str(int(year)-1)+'11']-df['apple_pay_purchase_amt'+str(int(year)-1)+'12']), df.match_apple_pay_purchase_amount_prev_3months_delta_2_1)
df['match_apple_pay_purchase_count_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_cnt'+str(int(year)-1)+'11']-df['apple_pay_purchase_cnt'+str(int(year)-1)+'12']), df.match_apple_pay_purchase_count_prev_3months_delta_2_1)
df['match_rewards_accrued_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['rewards_accrued'+str(int(year)-1)+'11']-df['rewards_accrued'+str(int(year)-1)+'12']), df.match_rewards_accrued_3months_delta_2_1)
elif feb:
if cohort_ind == 0:
df['match_aeroplan_expense_amount_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['aeroplan_expense_amt'+str(int(year)-1)+'11']+df['aeroplan_expense_amt'+year+'01']+df['aeroplan_expense_amt'+str(int(year)-1)+'12'])/3, 0)
df['match_apple_pay_purchase_amount_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_amt'+str(int(year)-1)+'11']+ df['apple_pay_purchase_amt'+year+'01']+ df['apple_pay_purchase_amt'+str(int(year)-1)+'12'])/3, 0)
df['match_apple_pay_purchase_count_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_cnt'+str(int(year)-1)+'11']+df['apple_pay_purchase_cnt'+year+'01']+df['apple_pay_purchase_cnt'+str(int(year)-1)+'12'])/3, 0)
df['match_rewards_accrued_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['rewards_accrued'+year+'01']+df['rewards_accrued'+str(int(year)-1)+'11']+df['rewards_accrued'+str(int(year)-1)+'12'])/3, 0)
df['match_aeroplan_expense_amount_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['aeroplan_expense_amt'+str(int(year)-1)+'11']-df['aeroplan_expense_amt'+str(int(year)-1)+'12']), 0)
df['match_apple_pay_purchase_amount_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_amt'+str(int(year)-1)+'11']-df['apple_pay_purchase_amt'+str(int(year)-1)+'12']), 0)
df['match_apple_pay_purchase_count_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_cnt'+str(int(year)-1)+'11']-df['apple_pay_purchase_cnt'+str(int(year)-1)+'12']), 0)
df['match_rewards_accrued_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['rewards_accrued'+str(int(year)-1)+'11']-df['rewards_accrued'+str(int(year)-1)+'12']), 0)
df['match_aeroplan_expense_amount_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['aeroplan_expense_amt'+str(int(year)-1)+'12']-df['aeroplan_expense_amt'+year+'01']), 0)
df['match_apple_pay_purchase_amount_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_amt'+str(int(year)-1)+'12']-df['apple_pay_purchase_amt'+year+'01']), 0)
df['match_apple_pay_purchase_count_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_cnt'+str(int(year)-1)+'12']-df['apple_pay_purchase_cnt'+year+'01']), 0)
df['match_rewards_accrued_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['rewards_accrued'+str(int(year)-1)+'12']-df['rewards_accrued'+year+'01']), 0)
else:
df['match_aeroplan_expense_amount_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['aeroplan_expense_amt'+str(int(year)-1)+'11']+df['aeroplan_expense_amt'+year+'01']+df['aeroplan_expense_amt'+str(int(year)-1)+'12'])/3, df.match_aeroplan_expense_amount_prev_3months)
df['match_apple_pay_purchase_amount_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_amt'+str(int(year)-1)+'11']+ df['apple_pay_purchase_amt'+year+'01']+ df['apple_pay_purchase_amt'+str(int(year)-1)+'12'])/3, df.match_apple_pay_purchase_amount_prev_3months)
df['match_apple_pay_purchase_count_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_cnt'+str(int(year)-1)+'11']+df['apple_pay_purchase_cnt'+year+'01']+df['apple_pay_purchase_cnt'+str(int(year)-1)+'12'])/3, df.match_apple_pay_purchase_count_prev_3months)
df['match_rewards_accrued_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['rewards_accrued'+year+'01']+df['rewards_accrued'+str(int(year)-1)+'11']+df['rewards_accrued'+str(int(year)-1)+'12'])/3, df.match_rewards_accrued_3months)
df['match_aeroplan_expense_amount_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['aeroplan_expense_amt'+str(int(year)-1)+'11']-df['aeroplan_expense_amt'+str(int(year)-1)+'12']), df.match_aeroplan_expense_amount_prev_3months_delta_3_2)
df['match_apple_pay_purchase_amount_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_amt'+str(int(year)-1)+'11']-df['apple_pay_purchase_amt'+str(int(year)-1)+'12']), df.match_apple_pay_purchase_amount_prev_3months_delta_3_2)
df['match_apple_pay_purchase_count_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_cnt'+str(int(year)-1)+'11']-df['apple_pay_purchase_cnt'+str(int(year)-1)+'12']), df.match_apple_pay_purchase_count_prev_3months_delta_3_2)
df['match_rewards_accrued_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['rewards_accrued'+str(int(year)-1)+'11']-df['rewards_accrued'+str(int(year)-1)+'12']), df.match_rewards_accrued_3months_delta_3_2)
df['match_aeroplan_expense_amount_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['aeroplan_expense_amt'+str(int(year)-1)+'12']-df['aeroplan_expense_amt'+year+'01']), df.match_aeroplan_expense_amount_prev_3months_delta_2_1)
df['match_apple_pay_purchase_amount_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_amt'+str(int(year)-1)+'12']-df['apple_pay_purchase_amt'+year+'01']), df.match_apple_pay_purchase_amount_prev_3months_delta_2_1)
df['match_apple_pay_purchase_count_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_cnt'+str(int(year)-1)+'12']-df['apple_pay_purchase_cnt'+year+'01']), df.match_apple_pay_purchase_count_prev_3months_delta_2_1)
df['match_rewards_accrued_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['rewards_accrued'+str(int(year)-1)+'12']-df['rewards_accrued'+year+'01']), df.match_rewards_accrued_3months_delta_2_1)
elif march:
if cohort_ind == 0:
df['match_aeroplan_expense_amount_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['aeroplan_expense_amt'+year+'02']+df['aeroplan_expense_amt'+year+'01']+df['aeroplan_expense_amt'+str(int(year)-1)+'12'])/3, 0)
df['match_apple_pay_purchase_amount_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_amt'+year+'02']+ df['apple_pay_purchase_amt'+year+'01']+ df['apple_pay_purchase_amt'+str(int(year)-1)+'12'])/3, 0)
df['match_apple_pay_purchase_count_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_cnt'+year+'02']+df['apple_pay_purchase_cnt'+year+'01']+df['apple_pay_purchase_cnt'+str(int(year)-1)+'12'])/3, 0)
df['match_rewards_accrued_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['rewards_accrued'+year+'02']+df['rewards_accrued'+year+'01']+df['rewards_accrued'+str(int(year)-1)+'12'])/3, 0)
df['match_aeroplan_expense_amount_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], -(df['aeroplan_expense_amt'+year+'01']-df['aeroplan_expense_amt'+str(int(year)-1)+'12']), 0)
df['match_apple_pay_purchase_amount_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], -(df['apple_pay_purchase_amt'+year+'01']- df['apple_pay_purchase_amt'+str(int(year)-1)+'12']), 0)
df['match_apple_pay_purchase_count_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], -(df['apple_pay_purchase_cnt'+year+'01']-df['apple_pay_purchase_cnt'+str(int(year)-1)+'12']), 0)
df['match_rewards_accrued_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], -(df['rewards_accrued'+year+'01']-df['rewards_accrued'+str(int(year)-1)+'12']), 0)
df['match_aeroplan_expense_amount_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['aeroplan_expense_amt'+year+'01']-df['aeroplan_expense_amt'+year+'02']), 0)
df['match_apple_pay_purchase_amount_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_amt'+year+'01']- df['apple_pay_purchase_amt'+year+'02']), 0)
df['match_apple_pay_purchase_count_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_cnt'+year+'01']-df['apple_pay_purchase_cnt'+year+'02']), 0)
df['match_rewards_accrued_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['rewards_accrued'+year+'01']-df['rewards_accrued'+year+'02']), 0)
else:
df['match_aeroplan_expense_amount_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['aeroplan_expense_amt'+year+'02']+df['aeroplan_expense_amt'+year+'01']+df['aeroplan_expense_amt'+str(int(year)-1)+'12'])/3, df.match_aeroplan_expense_amount_prev_3months)
df['match_apple_pay_purchase_amount_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_amt'+year+'02']+ df['apple_pay_purchase_amt'+year+'01']+ df['apple_pay_purchase_amt'+str(int(year)-1)+'12'])/3, df.match_apple_pay_purchase_amount_prev_3months)
df['match_apple_pay_purchase_count_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_cnt'+year+'02']+df['apple_pay_purchase_cnt'+year+'01']+df['apple_pay_purchase_cnt'+str(int(year)-1)+'12'])/3, df.match_apple_pay_purchase_count_prev_3months)
df['match_rewards_accrued_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['rewards_accrued'+year+'02']+df['rewards_accrued'+year+'01']+df['rewards_accrued'+str(int(year)-1)+'12'])/3, df.match_rewards_accrued_3months)
df['match_aeroplan_expense_amount_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], -(df['aeroplan_expense_amt'+year+'01']-df['aeroplan_expense_amt'+str(int(year)-1)+'12']), df.match_aeroplan_expense_amount_prev_3months_delta_3_2)
df['match_apple_pay_purchase_amount_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], -(df['apple_pay_purchase_amt'+year+'01']- df['apple_pay_purchase_amt'+str(int(year)-1)+'12']), df.match_apple_pay_purchase_amount_prev_3months_delta_3_2)
df['match_apple_pay_purchase_count_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], -(df['apple_pay_purchase_cnt'+year+'01']-df['apple_pay_purchase_cnt'+str(int(year)-1)+'12']), df.match_apple_pay_purchase_count_prev_3months_delta_3_2)
df['match_rewards_accrued_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], -(df['rewards_accrued'+year+'01']-df['rewards_accrued'+str(int(year)-1)+'12']), df.match_rewards_accrued_3months_delta_3_2)
df['match_aeroplan_expense_amount_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['aeroplan_expense_amt'+year+'01']-df['aeroplan_expense_amt'+year+'02']), df.match_aeroplan_expense_amount_prev_3months_delta_2_1)
df['match_apple_pay_purchase_amount_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_amt'+year+'01']- df['apple_pay_purchase_amt'+year+'02']), df.match_apple_pay_purchase_amount_prev_3months_delta_2_1)
df['match_apple_pay_purchase_count_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_cnt'+year+'01']-df['apple_pay_purchase_cnt'+year+'02']), df.match_apple_pay_purchase_count_prev_3months_delta_2_1)
df['match_rewards_accrued_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['rewards_accrued'+year+'01']-df['rewards_accrued'+year+'02']), df.match_rewards_accrued_3months_delta_2_1)
else:
if cohort_ind == 0:
df['match_aeroplan_expense_amount_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['aeroplan_expense_amt'+str(cohort_list[cohort_ind]-3)]+df['aeroplan_expense_amt'+str(cohort_list[cohort_ind]-1)]+df['aeroplan_expense_amt'+str(cohort_list[cohort_ind]-2)])/3, 0)
df['match_apple_pay_purchase_amount_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_amt'+str(cohort_list[cohort_ind]-3)]+ df['apple_pay_purchase_amt'+str(cohort_list[cohort_ind]-1)]+ df['apple_pay_purchase_amt'+str(cohort_list[cohort_ind]-2)])/3, 0)
df['match_apple_pay_purchase_count_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_cnt'+str(cohort_list[cohort_ind]-3)]+df['apple_pay_purchase_cnt'+str(cohort_list[cohort_ind]-1)]+df['apple_pay_purchase_cnt'+str(cohort_list[cohort_ind]-2)])/3, 0)
df['match_rewards_accrued_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['rewards_accrued'+str(cohort_list[cohort_ind]-3)]+df['rewards_accrued'+str(cohort_list[cohort_ind]-1)]+df['rewards_accrued'+str(cohort_list[cohort_ind]-2)])/3, 0)
df['match_aeroplan_expense_amount_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['aeroplan_expense_amt'+str(cohort_list[cohort_ind]-3)]-df['aeroplan_expense_amt'+str(cohort_list[cohort_ind]-2)]), 0)
df['match_apple_pay_purchase_amount_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_amt'+str(cohort_list[cohort_ind]-3)]-df['apple_pay_purchase_amt'+str(cohort_list[cohort_ind]-2)]), 0)
df['match_apple_pay_purchase_count_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_cnt'+str(cohort_list[cohort_ind]-3)]-df['apple_pay_purchase_cnt'+str(cohort_list[cohort_ind]-2)]), 0)
df['match_rewards_accrued_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['rewards_accrued'+str(cohort_list[cohort_ind]-3)]-df['rewards_accrued'+str(cohort_list[cohort_ind]-2)]), 0)
df['match_aeroplan_expense_amount_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['aeroplan_expense_amt'+str(cohort_list[cohort_ind]-2)]-df['aeroplan_expense_amt'+str(cohort_list[cohort_ind]-1)]), 0)
df['match_apple_pay_purchase_amount_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_amt'+str(cohort_list[cohort_ind]-2)]-df['apple_pay_purchase_amt'+str(cohort_list[cohort_ind]-1)]), 0)
df['match_apple_pay_purchase_count_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_cnt'+str(cohort_list[cohort_ind]-2)]-df['apple_pay_purchase_cnt'+str(cohort_list[cohort_ind]-1)]), 0)
df['match_rewards_accrued_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['rewards_accrued'+str(cohort_list[cohort_ind]-2)]-df['rewards_accrued'+str(cohort_list[cohort_ind]-1)]), 0)
else:
df['match_aeroplan_expense_amount_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['aeroplan_expense_amt'+str(cohort_list[cohort_ind]-3)]+df['aeroplan_expense_amt'+str(cohort_list[cohort_ind]-1)]+df['aeroplan_expense_amt'+str(cohort_list[cohort_ind]-2)])/3, df.match_aeroplan_expense_amount_prev_3months)
df['match_apple_pay_purchase_amount_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_amt'+str(cohort_list[cohort_ind]-3)]+df['apple_pay_purchase_amt'+str(cohort_list[cohort_ind]-1)]+df['apple_pay_purchase_amt'+str(cohort_list[cohort_ind]-2)])/3, df.match_apple_pay_purchase_amount_prev_3months)
df['match_apple_pay_purchase_count_prev_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_cnt'+str(cohort_list[cohort_ind]-3)]+df['apple_pay_purchase_cnt'+str(cohort_list[cohort_ind]-1)]+df['apple_pay_purchase_cnt'+str(cohort_list[cohort_ind]-2)])/3, df.match_apple_pay_purchase_count_prev_3months)
df['match_rewards_accrued_3months'] = np.where(df.cohort == cohort_list[cohort_ind], (df['rewards_accrued'+str(cohort_list[cohort_ind]-3)]+df['rewards_accrued'+str(cohort_list[cohort_ind]-1)]+df['rewards_accrued'+str(cohort_list[cohort_ind]-2)])/3, df.match_rewards_accrued_3months)
df['match_aeroplan_expense_amount_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['aeroplan_expense_amt'+str(cohort_list[cohort_ind]-3)]-df['aeroplan_expense_amt'+str(cohort_list[cohort_ind]-2)]), df.match_aeroplan_expense_amount_prev_3months_delta_3_2)
df['match_apple_pay_purchase_amount_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_amt'+str(cohort_list[cohort_ind]-3)]-df['apple_pay_purchase_amt'+str(cohort_list[cohort_ind]-2)]), df.match_apple_pay_purchase_amount_prev_3months_delta_3_2)
df['match_apple_pay_purchase_count_prev_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_cnt'+str(cohort_list[cohort_ind]-3)]-df['apple_pay_purchase_cnt'+str(cohort_list[cohort_ind]-2)]), df.match_apple_pay_purchase_count_prev_3months_delta_3_2)
df['match_rewards_accrued_3months_delta_3_2'] = np.where(df.cohort == cohort_list[cohort_ind], (df['rewards_accrued'+str(cohort_list[cohort_ind]-3)]-df['rewards_accrued'+str(cohort_list[cohort_ind]-2)]), df.match_rewards_accrued_3months_delta_3_2)
df['match_aeroplan_expense_amount_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['aeroplan_expense_amt'+str(cohort_list[cohort_ind]-2)]-df['aeroplan_expense_amt'+str(cohort_list[cohort_ind]-1)]), df.match_aeroplan_expense_amount_prev_3months_delta_2_1)
df['match_apple_pay_purchase_amount_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_amt'+str(cohort_list[cohort_ind]-2)]-df['apple_pay_purchase_amt'+str(cohort_list[cohort_ind]-1)]), df.match_apple_pay_purchase_amount_prev_3months_delta_2_1)
df['match_apple_pay_purchase_count_prev_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['apple_pay_purchase_cnt'+str(cohort_list[cohort_ind]-2)]-df['apple_pay_purchase_cnt'+str(cohort_list[cohort_ind]-1)]), df.match_apple_pay_purchase_count_prev_3months_delta_2_1)
df['match_rewards_accrued_3months_delta_2_1'] = np.where(df.cohort == cohort_list[cohort_ind], (df['rewards_accrued'+str(cohort_list[cohort_ind]-2)]-df['rewards_accrued'+str(cohort_list[cohort_ind]-1)]), df.match_rewards_accrued_3months_delta_2_1)
#set all the new columns to int32 instead of int64
df['match_aeroplan_expense_amount_prev_3months'] = df['match_aeroplan_expense_amount_prev_3months'].astype('int32')
df['match_apple_pay_purchase_amount_prev_3months'] = df['match_apple_pay_purchase_amount_prev_3months'].astype('int32')
df['match_apple_pay_purchase_count_prev_3months'] = df['match_apple_pay_purchase_count_prev_3months'].astype('int32')
df['match_rewards_accrued_3months'] = df['match_rewards_accrued_3months'].astype('int32')
df['match_aeroplan_expense_amount_prev_3months_delta_3_2'] = df['match_aeroplan_expense_amount_prev_3months_delta_3_2'].astype('float32')
df['match_apple_pay_purchase_amount_prev_3months_delta_3_2'] = df['match_apple_pay_purchase_amount_prev_3months_delta_3_2'].astype('float32')
df['match_apple_pay_purchase_count_prev_3months_delta_3_2'] = df['match_apple_pay_purchase_count_prev_3months_delta_3_2'].astype('float32')
df['match_rewards_accrued_3months_delta_3_2'] = df['match_rewards_accrued_3months_delta_3_2'].astype('float32')
df['match_aeroplan_expense_amount_prev_3months_delta_2_1'] = df['match_aeroplan_expense_amount_prev_3months_delta_2_1'].astype('float32')
df['match_apple_pay_purchase_amount_prev_3months_delta_2_1'] = df['match_apple_pay_purchase_amount_prev_3months_delta_2_1'].astype('float32')
df['match_apple_pay_purchase_count_prev_3months_delta_2_1'] = df['match_apple_pay_purchase_count_prev_3months_delta_2_1'].astype('float32')
df['match_rewards_accrued_3months_delta_2_1'] = df['match_rewards_accrued_3months_delta_2_1'].astype('float32')
#Remove index columns not required anymore
df = df.drop(aero_amt_list+apple_pay_amt_list+apple_pay_count_list+rewards_list, axis = 1)
return df
def transaction_acct(df):
'''
Add Count net Transactions, Net Transaction Amount, Travel Count, Travel Amt, Need Count, Need Amt, Want Cnt, Want Amt,
interchange, Pre Auth Cnt and Pre Auth Amt (Avg 3 months prior to move) to input dataframe
Input: Dataframe
Output: Dataframe
Effect: mutates input dataframe
'''
#init lists with columns from df
cnt_net_trans_list = list(df.filter(regex='cnt_net_transactions').columns)
net_trans_amt_list = list(df.filter(regex='sum_net_transactions_amt').columns)
trav_count_list = list(df.filter(regex='travel_cnt').columns)
trav_amt_list = list(df.filter(regex='travel_amt').columns)
need_cnt_list = list(df.filter(regex='need_cnt').columns)
need_amt_list = list(df.filter(regex='need_amt').columns)
want_count_list = list(df.filter(regex='want_cnt').columns)
want_amt_list = list(df.filter(regex='want_amt').columns)
interchange_list = list(df.filter(regex='interchange').columns)
preauth_amt_count_list = list(df.filter(regex='preauthorized_payment_amt').columns)