#分层随机抽样 stratified sampling

import xlrd, xlwt, time, random


xl = xlrd.open_workbook(r'C:\Users\Administrator\Desktop\分层抽样.xlsx')
xl_sht1 = xl.sheets()[0]
xl_sht1_nrows = xl_sht1.nrows
#表头
title = xl_sht1.row_values(0)
#把样本写进列表 sample
sample = []
for i in range(xl_sht1_nrows):
  sample.append(xl_sht1.row_values(i))
#打乱样本
random.shuffle(sample)
#把层的内容写进列表 col
col = xl_sht1.col_values(0)
#对col中的内容进行计数，获得每一类的名称对应个数的字典
col_dict = {}
for i in col:
  col_dict[i] = col_dict.get(i, 0) + 1
p = eval(input('每层抽取的比例（小数）：'))
#获得每一类的名称对应抽取个数的字典
col_p = {}
k = 0
for i in col_dict.keys():
  col_p[i] = int(round(col_dict[i] * p)) #round用来四舍五入，不加int结果会变成无数个p
#开始抽样，把抽取结果写进result_l列表
result_l = []
for i in sample:
  if col_p.get(i[0], 0) > 0:
    result_l.append(i)
    col_p[i[0]] -= 1
#新建文档，写入结果
f = xlwt.Workbook()
f_sht1 = f.add_sheet('result')
n = 0
for i in title:
  f_sht1.write(0, n, i)
  n += 1
r = 1
for i in result_l:
  c = 0
  for k in i:
    f_sht1.write(r, c, k)
    c += 1
  r += 1
nowtime = time.strftime("%Y%m%d_%H%M%S", time.localtime())
f.save(r'C:\Users\Administrator\Desktop\Result_{0}.xls'.format(nowtime))
print('\n抽样成功!')
input()

import pandas as pd
#对dataframe随机抽取2000个样本
pd.sample(df, n=2000)

from sklearn.model_selection import train_test_split
#y是在X中的某一个属性列
X_train, X_test, y_train, y_test = train_test_split(X,y, test_size=0.2, stratify=y)