Using np.select in DateTime series - Pandas Python

Question

I have a problem using np.select in DateTime series. Imagine you have a DataFrame: dateframe['id','FLOAT','DATE1','DATE2']

dateframe['DATE1'] = dt.datetime.now()
dateframe['DATE2'] = dt.datetime.now()

dateframe['DATE3'] = np.select(
    [
        dateframe['FLOAT']>0,
        dateframe['FLOAT']<=0,
    ],
    [
        dateframe['DATE1'],
        dateframe['DATE2'],
    ]
)

I get this error:

TypeError: The DTypes <class 'numpy.dtype[uint8]'> and <class 'numpy.dtype[datetime64]'> do not have a common DType. For example they cannot be stored in a single array unless the dtype is object.

Thanks in advance.

This is a data-specific issue, please post a [reproducible example](https://stackoverflow.com/questions/20109391/how-to-make-good-reproducible-pandas-examples). — Parfait, Jul 01 '21 at 22:44

David Erickson · Accepted Answer · 2021-07-01T22:37:08.213

use np.where when you have one condition that creates exactly two mutually exclusive groups:

dataframe['DATE1'] = dt.datetime.now()
dataframe['DATE2'] = dt.datetime.now()
dataframe['DATE3'] = np.where(dataframe['FLOAT']>0,dataframe['DATE1'],dataframe['DATE2'])
dataframe

Regarding the error with np.select. Initially store the date as an object datatype with str() and then convert to_datetime() later. Per the error, there is an issue with the datetime data type in the np.select statement

dataframe['DATE1'] = str(dt.datetime.now())
dataframe['DATE2'] = str(dt.datetime.now())
dataframe['DATE3'] = pd.to_datetime(np.select([(dataframe['FLOAT']>0),(dataframe['FLOAT']<=0)],
                                              [dataframe['DATE1'],dataframe['DATE2']]))
dataframe

Thanks a lot. It worked. I've looking for this answer for weeks. — Andres Perez, Jul 01 '21 at 22:52

Using np.select in DateTime series - Pandas Python

1 Answers1