题解 | #牛客网用户练习的平均次日留存率#

import pandas as pd
  
df = pd.read_csv('nowcoder.csv',sep=',',parse_dates=['date'])
df['date'] = df.date.dt.date.astype('datetime64[D]')
df = df.drop_duplicates(['user_id','date']).sort_values(['user_id','date']).reset_index(drop=True)
total = df['user_id'].count()
df.set_index('date',inplace=True)
res = df.groupby('user_id')['question_id'].rolling('2D').count()
num_act = res[res>=2].count()

print(f'{(num_act/total):.2f}')