题解 | #牛客网连续练习题目3天及以上的用户#

import pandas as pd
from datetime import timedelta
ar = pd.read_csv('nowcoder.csv',sep=',')
#转换日期格式+去重
ar.date=pd.to_datetime(ar.date).dt.date
ar.drop_duplicates(subset=['user_id','date'],keep='first',inplace=True)
#两次表格合并
c=pd.merge(ar,ar,on='user_id',suffixes=['_a','_b'])
d=pd.merge(c,ar,on='user_id')
#筛选
d_=d[((d.date_a+timedelta(days=1))==d.date_b)*((d.date_b+timedelta(days=1))==d.date)]
print(d_.user_id.groupby(d_.user_id).count())