yobx.sklearn.feature_selection.rfe

yobx.sklearn.feature_selection.rfe.sklearn_rfe(g: GraphBuilderExtendedProtocol, sts: Dict, outputs: List[str], estimator: RFE, X: str, name: str = 'rfe') → str

Converts a sklearn.feature_selection.RFE into ONNX.

After fitting, the transformer keeps the features selected by recursive feature elimination. The selected column indices are stored in estimator.get_support(indices=True) and are fixed constants at conversion time, so the ONNX graph contains a single Gather node that selects those columns from the input matrix:

X  ──Gather(axis=1, indices)──►  output

Parameters:

• g – the graph builder to add nodes to

• sts – shapes defined by scikit-learn

• estimator – a fitted RFE

• outputs – desired output names

• X – input name (shape (N, F))

• name – prefix name for the added nodes

Returns:

output name

yobx.sklearn.feature_selection.rfe.sklearn_rfecv(g: GraphBuilderExtendedProtocol, sts: Dict, outputs: List[str], estimator: RFECV, X: str, name: str = 'rfecv') → str

Converts a sklearn.feature_selection.RFECV into ONNX.

After fitting, the transformer keeps the features selected by recursive feature elimination with cross-validation. The selected column indices are stored in estimator.get_support(indices=True) and are fixed constants at conversion time, so the ONNX graph contains a single Gather node that selects those columns from the input matrix:

X  ──Gather(axis=1, indices)──►  output

Parameters:

• g – the graph builder to add nodes to

• sts – shapes defined by scikit-learn

• estimator – a fitted RFECV

• outputs – desired output names

• X – input name (shape (N, F))

• name – prefix name for the added nodes

Returns:

output name