在熊猫数据框中将整数转换为日期时间

Question

我用过 pd.to_numeric (df['dates'], downcast = 'integer') 将 datetime64 转换为 int64 但如果我想将其转换回 datetime 格式，我该怎么办？ 早些时候它的格式是：2020-02-19，但转换后现在是 1582070400000000000。我找不到将它恢复到 datetime64 的方法。

Answer 1

Pandas to_datatime 可以解决问题

参见示例：

import pandas as pd
# Example DataFrame
df = pd.DataFrame({'dates' : pd.date_range('01-01-2021','01-10-2021',freq='D')})
# To numeric, as you did
df['numeric'] = pd.to_numeric (df['dates'] , downcast = 'integer')
# Back to date
df['back_do_date'] = pd.to_datetime( df['numeric'] )

给出数据帧：df

    dates       numeric             back_do_date
0   2021-01-01  1609459200000000000 2021-01-01
1   2021-01-02  1609545600000000000 2021-01-02
2   2021-01-03  1609632000000000000 2021-01-03
3   2021-01-04  1609718400000000000 2021-01-04
4   2021-01-05  1609804800000000000 2021-01-05
5   2021-01-06  1609891200000000000 2021-01-06
6   2021-01-07  1609977600000000000 2021-01-07
7   2021-01-08  1610064000000000000 2021-01-08
8   2021-01-09  1610150400000000000 2021-01-09
9   2021-01-10  1610236800000000000 2021-01-10