python - 使用 OR 根据两列过滤数据框

Question

我需要过滤数据帧以减少更新用户属性的时间。

+----------+------------+------------+
| userCol1 |  dateCol1  |  dateCol2  |
+----------+------------+------------+
| user1    | 2020-01-16 | 2019-12-30 |
| user2    | 2019-10-31 | 2020-01-12 |
| user3    | 2019-08-15 | 2019-09-30 |
| user4    | 2019-08-25 | NaN        |
+----------+------------+------------+

上面是数据框的示例。我需要为 datecol1 的最新日期的任何用户过滤它或datecol2 is <= today-90 days 。在上面的例子中，上面的数据帧应该产生 user2和user4留在数据框中进行处理。

我编写的代码(尚未测试，所以我不知道它是否有效)不会过滤数据帧，而是尝试循环整个内容；这是代码。

 for row in df3.itertuples() :
     print(row.username)
     print(row.Password_Last_Set)
     print(row.Password_Last_forgot)
     if row.Password_Last_Forgot is 'NaN' and row.Password_Last_Set <= today.timedelta(days=90) :
         print('password expired based on last set, no forgot passwords')
     elif row.Password_Last_Forgot is not 'NaN' and row.Password_Last_Forgot > row.Password_Last_Set and row.Password_Last_Forgot <= today.timedelta(days=90) :
         print('password expired based on last forgot')
      elif row.Password_Last_Forgot is not 'NaN' and row.Password_Last_Forgot < row.Password_Last_Set and row.Password_Last_Set <= today.timedelta(days=90) :
         print('password expired based on last set')

如何在循环用户以对剩余用户执行操作之前进行过滤？

Answer 1

使用boolean indexing使用 max 表示最新日期时间:

df[['dateCol1','dateCol2']] = df[['dateCol1','dateCol2']].apply(pd.to_datetime)

cols = ['dateCol1','dateCol2']
df1 = df.loc[df[cols].max(axis=1)<=pd.Timestamp.now() - pd.Timedelta(90, unit='d'), 'userCol1']
print (df1)
2    user3
3    user4
Name: userCol1, dtype: object