分箱二

import sklearn
import numpy as np
import pandas as pd
import scipy
import matplotlib.pyplot as plt
plt.rcParams['font.sans-serif']=['SimHei'] #用来正常显示中文标签
plt.rcParams['axes.unicode_minus']=False #用来正常显示负号

model_data = pd.read_csv('model_data.csv',index_col=0)
model_data.head()

	SeriousDlqin2yrs	RevolvingUtilizationOfUnsecuredLines	age	NumberOfTime30-59DaysPastDueNotWorse	DebtRatio	MonthlyIncome	NumberOfOpenCreditLinesAndLoans	NumberOfTimes90DaysLate	NumberRealEstateLoansOrLines	NumberOfTime60-89DaysPastDueNotWorse	NumberOfDependents
0	0	0.015404	53.000000	0.000000	0.121802	4728.0	5.000000	0.0	0.000000	0.0	0.000000
1	0	0.168311	63.000000	0.000000	0.141964	1119.0	5.000000	0.0	0.000000	0.0	0.000000
2	1	1.063570	39.716057	1.141971	0.417663	3500.0	5.141971	1.0	0.141971	2.0	3.716057
3	0	0.088684	73.000000	0.000000	0.522822	5301.0	11.000000	0.0	2.000000	0.0	0.000000
4	1	0.622999	53.636002	0.000000	0.423650	13000.0	9.819992	0.0	2.000000	0.0	0.181999

auto_col_name = ['RevolvingUtilizationOfUnsecuredLines','age','DebtRatio','MonthlyIncome','NumberOfOpenCreditLinesAndLoans']

# 将分箱（一）中的代码封装为函数
def get_bins(data,col_name,max_bin_num,min_bin_num=2,target_col_name='SeriousDlqin2yrs',pic=True):
    try:
        bin_result,updown = pd.qcut(data[col_name],q=max_bin_num,retbins=True)
    except:
        max_bin_num -= 1
        bin_result,updown = pd.qcut(data[col_name],q=max_bin_num,retbins=True)
    bin_target = pd.DataFrame(bin_result)
    bin_target['target'] = model_data['SeriousDlqin2yrs']
    bin_target_count = bin_target.groupby(col_name)['target'].value_counts().unstack().fillna(0)
    target0_cnt = sum(bin_target_count[0])
    target1_cnt = sum(bin_target_count[1])
    box_info = [*zip(updown[:-1],updown[1:],bin_target_count[0],bin_target_count[1])]
                     #每组下限   每组上限    该年龄段0标签人数总计  该年龄段1标签人数总计
    
    # 保证每组中0 1标签的人数都不为0
    while sum(0 in box_info[i][2:] for i in range(len(box_info))): # 判断每个分组的两个标签的人数有没有0，如果有0就执行以下操作
        comb_index = [int(0 in box_info