riskcontrol 0.1.31

Added the calculation of GPS near geohash coding

Copyright (c) 2019 The Python Packaging Authority

The package is used for risk control modeling in Python. It mainly provides some basic and commonly data analysis methods for partners who want to learn Python data analysis or machine learning in the Internet financial industry. It also contains a large number of solutions to the problems encountered by the authors in their daily work. I hope you can actively use it. If you have any questions, please contact me at [e-mail].

riskcontrol

Riskcontrol is used for risk control modeling in Python.

It provides intuitive tools for

• feature information-value (iv)
• plot bad rate of bins based on decision tree
• ks plot
• feature info describe for external data validation(contain missing analysis)
• two feature heat map for cross analysis
• logistic credit card
• gps near geohash coding

install

pip install riskcontrol

Usage

import riskcontrol as rc
import pandas as pd

data = pd.read_csv('test.csv')
rc.feature_miss_ana(data) # data analysis report

col	col_type	iv	...	mode_v	mode_pect	mode_bad_rate	risk_monotonicity
ios	float64	0.0273	...	0	0.528996	0.054911	强
sex_m	float64	0.0486	...	1	0.723311	0.053726	强
age	float64	0.0408	...	22	0.095423	0.053391	弱
income	float64	0.0302	...	100000	0.147488	0.041681	弱
edu	float64	0.0367	...	0	0.828286	0.051514	强
freeram	float64	0.0087	...	482864	0.000104	0.125	无
joblevel	float64	0.0411	...	4	0.383198	0.056743	强

• miss_bad_rate:缺失值对应的bad_rate
• notnull_bad_rate：非缺失值对应的bad_rate
• zero_pect：0值占比
• zero_bad_rate：0值bad_rate
• mode_v：众数
• mode_pect：众数占比
• mode_bad_rate：众数对应的bad_rate
• risk_monotonicity：分箱后的逾期单调性

rc.split_box_plot_new(data, col_name) # bins bad rate plot

rc.create_geohash(latitude, longitude, radius, precision)