CODEWITHSUNDEEP
pythonpandasdatetime
Winds
Winds

Reputation: 397

python pandas datetime datetime does not work well

I have tried to solve this problem with a lot of methods but none of them worked.

I would like to convert a pandas series to datetime.

I think I may have miss some small but important part.

print test["order_date"][3]
print type(test["order_date"][3])

test["order_date"] = pd.to_datetime(test["order_date"], format="%Y-%m-%d %H:%M:%S")

##have also tried 
##test["order_date"] = pd.to_datetime(test["order_date"], infer_datetime_format=True) 
##test["order_date"] = test["order_date"].apply(pd.to_datetime)
##all turn out to be the same result

print test["order_date"][3]
print type(test["order_date"][3])

The outcome turns out to be like below:

20150731000001
<type 'numpy.int64'>

1970-01-01 05:35:50.731000001 
<class 'pandas.tslib.Timestamp'>

I cannot figure out why the result becomes 1970-01-01

If any further information is needed please kindly let me know.

Thanks!

Upvotes: 1

Views: 3569

Answers (2)

EdChum
EdChum

Reputation: 394051

It's because the dtype is int64 so it assumes nanosecond unit value:

In[51]:
pd.to_datetime(20150731000001, unit='ns')
Out[51]: Timestamp('1970-01-01 05:35:50.731000001')

If it was a string then it would be able to parse correctly:

In[54]:
pd.to_datetime('20150731000001')
Out[53]: Timestamp('2015-07-31 00:00:01')

So you can either explicitly pass a format string (like in @MaxU's answer) or convert the dtype of the column to str and then pass this:

test["order_date"] = pd.to_datetime(test["order_date"].astype(str))

Upvotes: 2

MaxU - stand with Ukraine
MaxU - stand with Ukraine

Reputation: 210842

You have specified a wrong format.

Try this:

In [34]: pd.to_datetime(['20150731000001'], format="%Y%m%d%H%M%S")
Out[34]: DatetimeIndex(['2015-07-31 00:00:01'], dtype='datetime64[ns]', freq=None)

Upvotes: 4

Related Questions