美团 python_Python | 美团差评数据分析

目录：数据理解

数据清洗

分析思路

分析过程

总结与建议

一.数据理解

数据来源于美团面试题，考点是分析差评原因，并给出改善方案。

字段理解：

数据大小：542 * 11

备注：骑手姓名重复默认为同一个骑手；同一个骑手可能在不同站点出现差评。

二.数据清洗

1.提取差评标签，转换成one-hot编码：

import numpy as np

import pandas as pd

import matplotlib.pyplot as plt

data = pd.read_excel('C:/Users/Hp/Desktop/temp/美团差评分析/不满意率分析0415.xlsx')

data = data.drop(columns='备注：\n1、分析差评原因，并给出改善方案；\n2、骑手姓名重复默认为同一个骑手；同一个骑手可能在不同站点出现差评\n')

data['评价标签'] = data['顾客配送评价标签'].str.split('|')

# 所有的评价标签

total_list = []

for i in data['评价标签']:

for j in i:

total_list.append(j)

set(total_list)

# 所有差评类型

# 判断是否仪表不整

def judge_1(x):

total = 0

for i in x['评价标签']:

if i == '仪表不整':

total += 1

return total

data['仪表不整'] = data.apply(judge_1, axis=1)

# 判断其他

def judge_2(x):

total = 0

for i in x['评价标签']:

if i == '其他':

total += 1

return total

data['其他'] = data.apply(judge_2, axis=1)

# 判断少餐/洒餐

def judge_3(x):

total = 0

for i in x['评价标签']:

if i == '少餐/洒餐':

total += 1

return total

data['少餐/洒餐'] = data.apply(judge_3, axis=1)

# 判断是否态度不好

def judge_4(x):

total = 0

for i in x['评价标签']:

if i == '态度不好':

total += 1

return total

data['态度不好'] = data.apply(judge_4, axis=1)

# 判断提前点送达

def judge_5(x):

total = 0

for i in x['评价标签']:

if i == '提前点送达':

total += 1

return total

data['提前点送达'] = data.apply(judge_5, axis=1)

# 判断送达不通知

def judge_6(x):

total = 0

for i in x['评价标签']:

if i == '送达不通知':

total += 1

return total

data['送达不通知'] = data.apply(judge_6, axis=1)

# 判断送达超时

def judge_7(x):

total = 0

for i in x['评价标签']:

if i == '送达超时':

total += 1