Can't convert dates to datetime64

Question

The following piece of code:

import pandas as pd
import numpy as np

data = pd.DataFrame({'date': ('13/02/2012', '14/02/2012')})
data['date'] = data['date'].astype('datetime64')

works fine on one machine (windows) and doesn't work on another (linux). Both numpy and pandas are installed on both.

The error I get is:

ValueError: Cannot create a NumPy datetime other than NaT with generic units

What does this error mean? I see it for the first time ever and there is not much on the web I can find. Is it some missing dependency?

Answer 1

Do this instead. Pandas keeps datestimes internally as datetime64[ns]. Conversions like this are very buggy (because of issues in various numpy version, 1.6.2 especially). Use the pandas routines, then operate like thesee are actual datetime objects. What are you trying to do?

In [30]: pandas.to_datetime(data['date'])
Out[30]: 
0   2012-02-13 00:00:00
1   2012-02-14 00:00:00
Name: date, dtype: datetime64[ns]