如何在每一天内使用最近的值来填充day？

Question

我有降水预报，但只预报早晨和晚上：

2022-06-08 00:00:00     NaN
2022-06-08 01:00:00     NaN
2022-06-08 02:00:00     NaN
2022-06-08 03:00:00     NaN
2022-06-08 04:00:00     NaN
2022-06-08 05:00:00     NaN
2022-06-08 06:00:00     NaN
2022-06-08 07:00:00    0.03
2022-06-08 08:00:00    0.03
2022-06-08 09:00:00    0.03
2022-06-08 10:00:00     NaN
2022-06-08 11:00:00     NaN
2022-06-08 12:00:00     NaN
2022-06-08 13:00:00     NaN
2022-06-08 14:00:00     NaN
2022-06-08 15:00:00     NaN
2022-06-08 16:00:00     NaN
2022-06-08 17:00:00     NaN
2022-06-08 18:00:00     NaN
2022-06-08 19:00:00    0.00
2022-06-08 20:00:00    0.00
2022-06-08 21:00:00    0.00
2022-06-08 22:00:00     NaN
2022-06-08 23:00:00     NaN
2022-06-09 00:00:00     NaN
2022-06-09 01:00:00     NaN
2022-06-09 02:00:00     NaN
2022-06-09 03:00:00     NaN
2022-06-09 04:00:00     NaN
2022-06-09 05:00:00     NaN
2022-06-09 06:00:00     NaN
2022-06-09 07:00:00    0.01

因为每个预报都是每日预报，所以我只想在每一天内填写。我可以

df['precipitation'] = df['precipitation'].groupby(pd.Grouper(freq = 'D')).ffill().bfill()

但这对早晨的天气预报影响过大，也低估了整个晚上的重量。我想做这样的事

df['precipitation'] = df['precipitation'].interpolate(method = 'nearest')

但按日期分组，这样天气预报就不会蔓延到另一天(这并不是预期的)。所需的输出类似于：

2022-06-08 00:00:00    0.03
2022-06-08 01:00:00    0.03
2022-06-08 02:00:00    0.03
2022-06-08 03:00:00    0.03
2022-06-08 04:00:00    0.03
2022-06-08 05:00:00    0.03
2022-06-08 06:00:00    0.03
2022-06-08 07:00:00    0.03
2022-06-08 08:00:00    0.03
2022-06-08 09:00:00    0.03
2022-06-08 10:00:00    0.03
2022-06-08 11:00:00    0.03
2022-06-08 12:00:00    0.03
2022-06-08 13:00:00    0.03
2022-06-08 14:00:00    0.00
2022-06-08 15:00:00    0.00
2022-06-08 16:00:00    0.00
2022-06-08 17:00:00    0.00
2022-06-08 18:00:00    0.00
2022-06-08 19:00:00    0.00
2022-06-08 20:00:00    0.00
2022-06-08 21:00:00    0.00
2022-06-08 22:00:00    0.00
2022-06-08 23:00:00    0.00
2022-06-09 00:00:00    0.01
2022-06-09 01:00:00    0.01
2022-06-09 02:00:00    0.01
2022-06-09 03:00:00    0.01
2022-06-09 04:00:00    0.01
2022-06-09 05:00:00    0.01
2022-06-09 06:00:00    0.01
2022-06-09 07:00:00    0.01

Answer 1

您可以简单地将插值应用于您的组：

df.groupby(pd.Grouper(freq = 'D')).apply(pd.DataFrame.interpolate, method = 'nearest')

或者使用lambda函数

df.groupby(pd.Grouper(freq = 'D')).apply(lambda group: group.interpolate(method = 'nearest'))

请注意，在大型数据文件上的表现可能会很糟糕。