bvslusa 0.0.2

Functions to work with BVS enriched data

bvslusa

Install

pip install bvslusa

How to use

import pandas as pd
from bvslusa.validate import remove_restritivos, target_mapping
from bvslusa.evaluate import evaluate_bvs_scores
from bvslusa.ratings import get_ratings

df = pd.read_csv('../data/AVZA_FB727003.csv', sep=';')
df

<style scoped> .dataframe tbody tr th:only-of-type { vertical-align: middle; } .dataframe tbody tr th { vertical-align: top; } .dataframe thead th { text-align: right; } </style>

	DOC_NUMBER	SAFRA	QTD_SCPC	VL_SCPC	QTD_CCF	QTD_PROTESTO	VL_PROTESTO	FLAG_RESTRITIVO	SCRCRDMERPJ3	SCRCRDMERPJ4	SCRCRDMERPJ5	SCRCRDATACAD	SCRCRDMERMEI	PERF_MERC_60D6M_EVER
0	891026	202203	0	0	1	0	0	1	267	256	58	185	5	MAU
1	982383	202203	0	0	0	0	0	0	283	256	460	619	30	BOM
2	176129	202203	0	0	0	0	0	0	283	256	460	712	34	BOM
3	566081	202203	0	0	0	0	0	0	283	256	460	781	30	BOM
4	760613	202203	0	0	0	0	0	0	283	256	460	712	34	BOM
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
35734	776859	202203	0	0	0	4	911	1	484	767	513	969	902	MAU
35735	94325	202203	0	0	0	4	911	1	484	767	513	969	902	MAU
35736	315930	202203	0	0	0	4	911	1	484	767	513	969	902	MAU
35737	668323	202203	0	0	0	4	911	1	484	767	513	969	902	MAU
35738	140483	202203	0	0	0	2	5285	1	487	767	531	956	131	BOM

35739 rows × 14 columns

df = remove_restritivos(df)
df = target_mapping(df, target='PERF_MERC_60D6M_EVER', map_dict={'BOM': 0, 'MAU': 1})
df.pipe(evaluate_bvs_scores)

<style scoped> .dataframe tbody tr th:only-of-type { vertical-align: middle; } .dataframe tbody tr th { vertical-align: top; } .dataframe thead th { text-align: right; } </style>

	score	auc	ks
0	SCRCRDMERPJ5	0.694770	30.86
1	SCRCRDATACAD	0.666147	28.28
2	SCRCRDMERMEI	0.610453	19.22
3	SCRCRDMERPJ3	0.598126	16.93
4	SCRCRDMERPJ4	0.574876	13.56

As the SCRCRDMERPJ5 is the best bvs score, let’s proceed to build ratings with it:

df_ratings = get_ratings(df, target='PERF_MERC_60D6M_EVER', score='SCRCRDMERPJ5')

<style scoped> .dataframe tbody tr th:only-of-type { vertical-align: middle; } .dataframe tbody tr th { vertical-align: top; } .dataframe thead th { text-align: right; } </style>

	Bin	Count	Count (%)	Non-event	Event	Event rate	WoE	IV	JS
0	(-inf, 294.50)	3341	0.142900	2709	632	0.189165	-0.892188	0.163145	1.974255e-02
1	[294.50, 357.50)	1336	0.057143	1107	229	0.171407	-0.771947	0.046611	5.685828e-03
2	[357.50, 405.50)	1380	0.059025	1184	196	0.142029	-0.549094	0.022291	2.751878e-03
3	[405.50, 457.50)	1469	0.062831	1299	170	0.115725	-0.314082	0.007055	8.782459e-04
4	[457.50, 491.50)	1406	0.060137	1275	131	0.093172	-0.072129	0.000322	4.028209e-05
5	[491.50, 535.50)	1547	0.066168	1413	134	0.086619	0.007997	0.000004	5.272191e-07
6	[535.50, 570.50)	1189	0.050855	1095	94	0.079058	0.107581	0.000563	7.034663e-05
7	[570.50, 597.50)	1276	0.054577	1200	76	0.059561	0.41171	0.007813	9.698149e-04
8	[597.50, 629.50)	1345	0.057528	1269	76	0.056506	0.467618	0.010386	1.286575e-03
9	[629.50, 682.50)	2340	0.100086	2223	117	0.050000	0.596806	0.027941	3.441700e-03
10	[682.50, 748.50)	3183	0.136142	3068	115	0.036129	0.936216	0.081821	9.869692e-03
11	[748.50, inf)	3568	0.152609	3498	70	0.019619	1.563818	0.202677	2.303251e-02
12	Special	0	0.000000	0	0	0.000000	0.0	0.000000	0.000000e+00
13	Missing	0	0.000000	0	0	0.000000	0.0	0.000000	0.000000e+00
Totals		23380	1.000000	21340	2040	0.087254		0.570628	6.776995e-02