python的等深分箱实例

# -*- coding: utf-8 -*-
"""
Created on Tue Jan 29 17:26:38 2019
@author: DamomWCG
"""
 
 
 
class Equal_depth_box:
  
 def equal_box(list, bin_num):
  '''
  param:
  list:you need bin box list
  bin_num: you want bin num
  '''
  bin_num = 10
  list.sort() #need sort can replace by others
  list_2 = list.copy()
  all_num = len(list_2)
  bin_sep = all_num/bin_num
  bin_sep = int(bin_sep)
  bin_list = []
  for i in range(1,bin_num):
   bin_dict = {}
   bin_dict = i*bin_sep
   bin_list.append(bin_dict)
  bin_real_list = []
  for i in bin_list:
   #print(i)
   bin_real_dict = {}
   bin_real_dict = list[i]
   bin_real_list.append(bin_real_dict)
  
  return bin_real_list
  
 def replace_box(list_1,replace_list):
  '''
  param:
  list_1:you need bin list
  replace_list: from equal box, replace the original list
  '''
  import pandas as pd
  list_max = max(list_1)
  list_min = min(list_1)
  replace_list.insert(0,list_min -1) #insert start
  replace_list.append(list_max + 1) #insert end
 
  list_2 = pd.cut(list_1, bins = replace_list,
      labels = range(len(replace_list) - 1)).tolist()
  return list_2
  
  
  
 
 
if __name__ == '__main__':
 
  import random
  
  list_1 = random.sample(range(1000), 134)
  
  print(list_1.sort())
  
  print('real_list: {}'.format(list_1[0:50]))
  
  replace_list = bin_class.equal_box(list_1, 10)
 
  list_2 = bin_class.replace_box(list_1, replace_list)
  
  print('encode_list: {}'.format(list_2[0:50]))

# -*- coding: utf-8 -*-
"""
Created on Wed Jan 30 16:17:19 2019
@author: DamonWCG
"""
 
from Equal_depth_box import *
 
import pandas as pd
 
df = pd.read_csv('test.csv', encoding = 'gbk')
 
df.columns
####需要分箱的列
list_1 = df['deal_city_encoding'].tolist() #本方法是针对于list，所以对于series需要进行变换
 
####需要分箱的个数
replace_list = Equal_depth_box.equal_box(list_1, 5) 
 
 
####替代的名称
 
##因为列表排序所以需要重新排序对齐，这里我有空再想想其他办法
df.sort_values(by="deal_city_encoding", inplace = True)
 
list_2 = Equal_depth_box.replace_box(list_1, replace_list)
 
df['deal_city_bin_encoding'] = list_2
 
 
 
df.to_csv('df.csv', encoding = 'gbk', index = False)