如何将 pandas dataframe 列中的两种不同日期格式转换为相同格式？

Question

I have two different date formats in a pandas column such as - DD-MM-YYYY and MM/DD/YYYY and I want to convert them into the same format.

I tried using the code -

data['SALE DATE'] = pd.to_datetime(data['SALE DATE']).dt.strftime('%m/%d/%Y')

but this converts the dates into DD/MM/YYYY and MM/DD/YYYY into the output - data['SALE DATE']

I want a python solution to overcome this problem. Any leads will be very helpful.

Answer 1

The most intuitive solution is to write a custom conversion function, someting like:

def myDateConv(tt):
    sep = tt[2]
    if sep == '-':
        return pd.to_datetime(tt, format='%d-%m-%Y')
    elif sep == '/':
        return pd.to_datetime(tt, format='%m/%d/%Y')
    else:
        return tt

and then pass it as a converter for the column in question:

df = pd.read_csv('Input.csv', converters={'Date': myDateConv})

I prepared a CSV file, which read with read_csv without any custom converter gave the original content and both columns of object type:

         Date Input format
0  03-05-2020   DD-MM-YYYY
1  05/07/2020   MM/DD/YYYY

But reading the same file with the above converter gave:

        Date Input format
0 2020-05-03   DD-MM-YYYY
1 2020-05-07   MM/DD/YYYY

with Date column of datetime64[ns] type and both dates from May, just as intended.

Or if you have this DataFrame from other source and you want to convert this column, run:

df.Date = df.Date.apply(myDateConv)

Answer 2

If you are using pandas version 1.xx you can use the following solution:

pd.to_datetime(["11-08-2018", "05-03-2016", "08/30/2017", "09/21/2018"], infer_datetime_format=True, dayfirst=True).strftime("%m/%d/%Y")

This gives the following result:

Index(['08/11/2018', '03/05/2016', '08/30/2017', '09/21/2018'], dtype='object')

... the important argument here is dayfirst=True .

See pd.to_datetime docs for more.