snorkelflow.client.analyses.plot_feature_importance

snorkelflow.client.analyses.plot_feature_importance(node, model_uid, split='valid')

Generates a SHAP based summary plot.

note

This function requires these additional dependencies to be installed: shap, transformers, xgboost. They are pre-installed in the in-platform notebook server but may not be present in your own Python environment. Please make sure to have them installed. We’ve tested with shap==0.40.0, transformers==4.27.4, xgboost==1.1.0.

Parameters

Parameters

Name	Type	Default	Info
node	int		UID of the node.
model_uid	int		UID of the model to get important features for.
split	str	'valid'	The split to get a visualization for.

Raises

Raises

NotSupportedException – For models other than LogReg and XGBoost.

Return type

Return type

None