问题
datetime
2012-01-01 125.5010
2012-01-02 NaN
2012-01-03 125.5010
2013-01-04 NaN
2013-01-05 125.5010
2013-02-28 125.5010
2014-02-28 125.5010
2016-01-02 125.5010
2016-01-04 125.5010
2016-02-28 NaN
I would like to fill in the missig values in this dataframe by using a climatology computed from the dataset i.e fill in missing 28th feb 2016 value by averaging values of 28th feb from other years. How do i do this?
回答1:
You can use groupby by month and day and transform with fillna mean:
print df.groupby([df.index.month, df.index.day]).transform(lambda x: x.fillna(x.mean()))
datetime
2012-01-01 125.501
2012-01-02 125.501
2012-01-03 125.501
2013-01-04 125.501
2013-01-05 125.501
2013-02-28 125.501
2014-02-28 125.501
2016-01-02 125.501
2016-01-04 125.501
2016-02-28 125.501
来源：https://stackoverflow.com/questions/34972297/fill-in-missing-values-in-pandas-dataframe-using-mean