how to process all the columns of pandas dataframe