How to downcast numeric columns in Pandas?

How to optimize the data frame memory footprint and find the most optimal (minimal) data types dtypes for numeric columns. For example:

   A        B    C         D
0  1  1000000  1.1  1.111111
1  2 -1000000  2.1  2.111111

>>> df.dtypes
A      int64
B      int64
C    float64
D    float64

Expected result:

>>> df.dtypes
A       int8
B      int32
C    float32
D    float32
dtype: object

Answer

You can use parameter downcast in to_numeric with selectig integers and floats columns by DataFrame.select_dtypes, it working from pandas 0.19+ like mentioned @anurag, thank you:

fcols = df.select_dtypes('float').columns
icols = df.select_dtypes('integer').columns

df[fcols] = df[fcols].apply(pd.to_numeric, downcast='float')
df[icols] = df[icols].apply(pd.to_numeric, downcast='integer')

print (df.dtypes)
A       int8
B      int32
C    float32
D    float32
dtype: object