Cross Validation

Cross-validation utilities.

diffupath.repeated_holdout.validation_by_method(mapping_input: Union[List, Dict[str, List]], graph: networkx.classes.graph.Graph, kernel: diffupy.matrix.Matrix, k: Optional[int] = 100) → Tuple[Dict[str, list], Dict[str, list]]

Repeated holdout validation by diffustion method.

Parameters

• mapping_input – List or value dictionary of labels {‘label’:value}.

• graph – Network as a graph object.

• kernel – Network as a kernel.

• k – Iterations for the repeated_holdout validation.

diffupath.repeated_holdout.validation_by_subgraph(mapping_input, kernels: Dict[str, List[diffupy.matrix.Matrix]], universe_kernel: Optional[diffupy.matrix.Matrix] = None, z_normalization: Optional[bool] = True, k: Optional[int] = 100) → Tuple[Dict[str, Dict[str, List]], Dict[str, Dict[str, List]]]

Repeated holdout validation by subgraph.

Parameters

• mapping_input – List or value dictionary of labels {‘label’:value}.

• kernels – Network stratified as a dictionary {‘kernel-tile’:kernel}.

• universe_kernel – Network as an integrated kernel.

• z_normalization – Flag for the statistical normalization option.

• k – Iterations for the repeated_holdout validation.