Convertir un tableau d'objets NumPy en datetime64

Question

data['RealTime'][:,0] 
Out[23]: 
array([datetime.datetime(2017, 9, 12, 18, 13, 8, 826000), 
     datetime.datetime(2017, 9, 12, 18, 13, 8, 846000), 
     datetime.datetime(2017, 9, 12, 18, 13, 8, 866000), ..., 
     datetime.datetime(2017, 9, 12, 18, 30, 40, 186000), 
     datetime.datetime(2017, 9, 12, 18, 30, 40, 206000), 
     datetime.datetime(2017, 9, 12, 18, 30, 40, 226000)], dtype=object) 

Comment puis-je convertir en un tableau de datetime dtype?

Answer 1

Je sais que vous avez pandas, donc vous pouvez simplement utiliser pd.to_datetime:

out = pd.to_datetime(array) 
print(out) 

DatetimeIndex(['2017-09-12 18:13:08.826000', '2017-09-12 18:13:08.846000', 
       '2017-09-12 18:13:08.866000', '2017-09-12 18:30:40.186000', 
       '2017-09-12 18:30:40.206000', '2017-09-12 18:30:40.226000'], 
       dtype='datetime64[ns]', freq=None) 

Vous pouvez récupérer un tableau numpy de out en accédant out.values.

---

Avec numpy, vous feriez la même chose en utilisant astype:

out = array.astype("datetime64[ns]") 
print(out) 

array(['2017-09-12T18:13:08.826000000', '2017-09-12T18:13:08.846000000', 
     '2017-09-12T18:13:08.866000000', '2017-09-12T18:30:40.186000000', 
     '2017-09-12T18:30:40.206000000', '2017-09-12T18:30:40.226000000'], dtype='datetime64[ns]')