Column Selection: Compatibility with sklearn.compose.make_column_selector

[mdancho84]

Hey, awesome package. I'm planning to use it in my python course I'm developing.

Question

Is it possible to apply transformations by data type or with a column selection method similar to sklearn.compose.make_column_selector?

Example

A simple example is to select all numeric columns for scaling. With sklearn I'd do something like this:

column_transformer_scale_numeric = make_column_transformer(
    (StandardScaler(), make_column_selector(dtype_include = np.number)), 
    remainder = 'passthrough'
)

Ideally I'd be able to use the make_column_selector() from sklearn and apply it within the sklearn_pandas.DataFrameMapper().

mapper_scale = DataFrameMapper(
    features  = [
        (make_column_selector(dtype_include = np.number), StandardScaler())
    ]
)

Any thoughts on this? Also let me know if this has been addressed previously and if I just missed searching for it.

Thanks!

[ragrawal]

Hi @mdancho84 , no this is not yet implemented. The ideas was previously discussed but we never got it implemented. Let me think about it. One workaround for now is to use gen_feature as shown in README. Let me know if that works for you