CODEWITHSUNDEEP
pythonpandasdatetime
deblue
deblue

Reputation: 267

Set up a DataFrame index to be DateTime

I have a pd.DataFrame which has datetime64 type of index. I'd like to concatenate a pd.Series column to it and map it to the index.

However, my pd.Series has an object type of dates, while my pd.DataFrame has a datetime64 type of dates.

I tried converting the pd.Series index to datetime like so.

This is how the pd.Series looks like :

west_index[1:5]
Out[40]: 
                 WEST
Exchange Date        
2005-01-05    -0.7466
2005-01-06        NaN
2005-01-07     0.3401
2005-01-10    -0.0688

I did:

a = [datetime.strptime(_, '%Y-%m-%d') for _ in west_index.index]
west_index.set_index(a, inplace=True)

When I try to concatenate to my main pd.DataFrame, which looks like this:

log_r[500:505]
Out[42]: 
                BELEX15       BET       BUX  ...    STOXX50     WIG20  
Exchange Date                                ...                           
2006-12-08     0.131916  1.717272 -0.983922  ...   0.090416 -0.436533   
2006-12-11     0.191401 -0.645286  1.158256  ...   0.576408 -0.086710   
2006-12-12    -0.190785 -0.764530  0.753001  ...   0.396716 -0.684809   
2006-12-13    -0.521898  0.726442 -1.459122  ...   0.629700  0.202408   
2006-12-14     0.877326 -0.707257  1.099898  ...   0.718143  0.722291   


log_r = pd.concat([log_r,west_index], axis=1)

I get an error

KeyError: datetime.datetime(2005, 1, 4, 0, 0)

Upvotes: 0

Views: 103

Answers (1)

jezrael
jezrael

Reputation: 862406

I think better is convert log_r.index or west_index.index to datetimes for datetimes in both indexes by to_datetime:

#check it
print (west_index.index.dtype)
print (log_r.index.dtype)

west_index.index = pd.to_datetime(west_index.index)
log_r.index = pd.to_datetime(log_r.index)
log_r = pd.concat([log_r, west_index], axis=1)

Upvotes: 1

Related Questions