将 float64 列转换为日期时间熊猫时出错

Question

I need to convert float64 type to datetime format.

As an example 20181219.0 data, I want is as 2018-12-19

I have tried common code;

df1['ACT_DATE1'] = pd.to_datetime(df1['ACT_DATE1'], format='%Y/%m/%d')

But, what I get is 1970-01-01 00:00:00.020181219

Answer 1

First obviously problem here are some missing values, which are converted to NaT for missing datetimes.

Tested in pandas 0.25.0 and there is no problem with .0 values only necessary specify format of datetimes - %Y%m%d :

df1 = pd.DataFrame({'ACT_DATE1' : [20181219.0, np.nan]})

df1['ACT_DATE1'] = pd.to_datetime(df1['ACT_DATE1'], format='%Y%m%d') 
print (df1)
   ACT_DATE1
0 2018-12-19
1        NaT

If not working and is necessary remove missing values and converting to integer:

dates = df1['ACT_DATE1'].dropna().astype(int)
df1['ACT_DATE1'] = pd.to_datetime(dates, format='%Y%m%d')