Python Pandas: convert <class 'tuple'> to datetime

Question

I have a DataFrame looking like this

         Sigma     JulianDay  
 0       -9.05  2.451545e+06 
 1      -10.99  2.451545e+06 
 2       -8.42  2.451546e+06 
 3       -8.92  2.451547e+06 
 4      -10.79  2.451547e+06
 5       -9.53  2.451548e+06

I want to convert the column 'JulianDay' to Gregorian Date. For that, I used the jdcal 1.3 package with the already defined function jd_to_date(jd)

d = df['JulianDay'].apply(jd_to_date)
df['GregDate'] = d

and the result is:

        Sigma     JulianDay       GregDate
0       -9.05  2.451545e+06   (2000, 1, 1)
1      -10.99  2.451545e+06   (2000, 1, 1) 
2       -8.42  2.451546e+06   (2000, 1, 2)
3       -8.92  2.451547e+06   (2000, 1, 3)
4      -10.79  2.451547e+06   (2000, 1, 3)
5       -9.53  2.451548e+06   (2000, 1, 4)

The output also tells me, that I am dealing with an object type:

Name: JulianDay, dtype: object

My problem is now, that I want to convert the GregDate column to datetime, so that I am able to set the GregDate as index. But the dtype: object gives me trouble when trying to

df['GregDate'] = pd.to_datetime(d)

I get the error:

TypeError: <class 'tuple'> is not convertible to datetime

So, how do I first convert the "tuple" column to something else, so that I am able to convert it to datetime?

Thanks

Answer 1

You can create new DataFrame with constructor and then apply to_datetime, important are column names year, month and day:

a = pd.DataFrame(df['GregDate'].values.tolist(), columns=['year','month','day'])
print (a)
   year  month  day
0  2000      1    1
1  2000      1    1
2  2000      1    2
3  2000      1    3
4  2000      1    3
5  2000      1    4

df.GregDate = pd.to_datetime(a)
print (df)
   Sigma  JulianDay   GregDate
0  -9.05  2451545.0 2000-01-01
1 -10.99  2451545.0 2000-01-01
2  -8.42  2451546.0 2000-01-02
3  -8.92  2451547.0 2000-01-03
4 -10.79  2451547.0 2000-01-03
5  -9.53  2451548.0 2000-01-04