How to convert Dataframe containing float and nan values to datetime python?

I have a dataframe float column as:

data = {'mydate': [23131.0,23131.0,np.nan,22677.0,22554.0,np.nan,23131.0]}    df = pd.DataFrame(data,columns=['mydate'])             mydate   0        23131.0                1        23131.0                2        NaN                3        22677.0                4        22554.0                5        NaN                6        23131.0                

It contains null values. I am trying to convert it to datetime python using the following code

def dayym(unit):      dates = {date:((epoch + datetime.timedelta(days=date))) for date in unit.unique()}      return unit.map(dates)    df.loc[:,'mydate']= dayym(df['mydate'])  

with the following error:

    dates = {date:((epoch + datetime.timedelta(days=date))) for date in unit.unique()}    File "central_read.py", line 18, in <dictcomp>      dates = {date:((epoch + datetime.timedelta(days=date))) for date in unit.unique()}  ValueError: cannot convert float NaN to integer  

Any ideas. I am out of them at this point.

https://stackoverflow.com/questions/66913586/how-to-convert-dataframe-containing-float-and-nan-values-to-datetime-python April 02, 2021 at 09:56AM