df.astype({'col1': 'int32'}).dtypes col1 int32 col2 int64 dtype: object
df.astype('int64').dtypes #www.athulmathew.com