I need to use different functions to treat numeric columns and string columns. What I am doing now is really dumb: allc = list((agg.loc[:, (agg.dtypes==np.float64)|(agg.dtypes==np.int)]).columns) ...