How to use Timestamp Data in Building Naive Bayes model in Python

Question

I have a Dataset, with Timestamp as one of the column with the format 09/07/2016 23:58.

I'm trying to apply Naive Bayes on this Data, and i'm facing the below error. Please let me know how to use this Data in my model

ValueError: invalid literal for float(): 12/06/2016 23:59

Answer 1

You need to_datetime with parameter errors='coerce' for convert bad not parseable values to NaT:

df = pd.DataFrame({'date':['12/06/2016 23:59','12/06/2016 23:59', 'a']})
print (df)
               date
0  12/06/2016 23:59
1  12/06/2016 23:59
2                 a


print (pd.to_datetime(df.date, errors='coerce'))
0   2016-12-06 23:59:00
1   2016-12-06 23:59:00
2                   NaT
Name: date, dtype: datetime64[ns]

For test bad values use boolean indexing - return all rows where is NaT:

print (df[pd.to_datetime(df.date, errors='coerce').isnull()])
  date
2    a