sdkagent

shap API reference

76 public APIs from shap (shap/shap) — 28 classes, 27 functions, 21 methods. Signatures extracted by static analysis of the actual source.

Repository: shap/shap

KindCount
Classes28
Functions27
Methods21

API list

classshap._explanation.Explanation
A sliceable set of parallel arrays representing a SHAP explanation.
methodshap._explanation.Explanation.cohorts(cohorts:int | list[int] | tuple[int] | np.ndarray) -> Cohorts
Split this explanation into several cohorts.
methodshap._explanation.Explanation.hstack(other:Explanation) -> Explanation
Stack two explanations column-wise.
methodshap._explanation.Explanation.max(axis:int) -> Explanation
Numpy-style max function.
methodshap._explanation.Explanation.mean(axis:int) -> Explanation
Numpy-style mean function.
methodshap._explanation.Explanation.min(axis:int) -> Explanation
Numpy-style min function.
methodshap._explanation.Explanation.shape() -> tuple[int, ...]
Compute the shape over potentially complex data nesting.
methodshap._explanation.Explanation.sum(axis:int | None=None, grouping:dict[str, str] | None=None) -> Explanation
Numpy-style sum function.
methodshap._explanation.Explanation.values()
Pass-through from the underlying slicer object.
classshap._explanation.OpHistoryItem
An operation that has been applied to an Explanation object.
classshap._serializable.Deserializer
Load data items from an input stream.
classshap._serializable.Serializable
This is the superclass of all serializable objects.
methodshap._serializable.Serializable.save(out_file)
Save the model to the given file stream.
classshap._serializable.Serializer
Save data items to an input stream.
classshap.actions._action.Action
Abstract action class.
funcshap.datasets.a1a(n_points:int | None=None) -> tuple[ssp.csr_matrix, np.ndarray]
Return a sparse dataset in scipy csr matrix format.
funcshap.datasets.adult(display:bool=False, n_points:int | None=None) -> tuple[pd.DataFrame, np.ndarray]
Return the Adult census data in a structured format.
funcshap.datasets.cache(url:str, file_name:str | None=None) -> str
Loads a file from the URL and caches it locally.
funcshap.datasets.california(n_points:int | None=None) -> tuple[pd.DataFrame, np.ndarray]
Return the California housing data in a tabular format.
funcshap.datasets.diabetes(n_points:int | None=None) -> tuple[pd.DataFrame, np.ndarray]
Return the diabetes data in a nice package.
funcshap.datasets.imagenet50(resolution:int=224, n_points:int | None=None) -> tuple[np.ndarray, np.ndarray]
Return a set of 50 images representative of ImageNet images.
classshap.explainers._additive.AdditiveExplainer
Computes SHAP values for generalized additive models.
methodshap.explainers._additive.AdditiveExplainer.supports_model_with_masker(model:Any, masker:Any) -> bool
Determines if this explainer can handle the given model.
classshap.explainers._deep.DeepExplainer
Meant to approximate SHAP values for deep learning models.
funcshap.explainers._deep.deep_pytorch.linear_1d(module, grad_input, grad_output)
No change made to gradients.
funcshap.explainers._deep.deep_pytorch.passthrough(module, grad_input, grad_output)
No change made to gradients
classshap.explainers._exact.ExactExplainer
Computes SHAP values via an optimized exact enumeration.
funcshap.explainers._exact.gray_code_indexes(nbits:int) -> npt.NDArray[np.intp]
Produces an array of which bits flip at which position.
classshap.explainers._gpu_tree.GPUTreeExplainer
Experimental GPU accelerated version of TreeExplainer.
classshap.explainers._tree.SingleTree
A single decision tree.
classshap.explainers._tree.TreeEnsemble
An ensemble of decision trees.
methodshap.explainers._tree.TreeEnsemble.get_transform() -> str
A consistent interface to make predictions from this model.
methodshap.explainers._tree.TreeExplainer.supports_model_with_masker(model:Any, masker:Any) -> bool
Determines if this explainer can handle the given model.
classshap.explainers._tree.XGBTreeModelLoader
This loads an XGBoost model directly from a raw memory dump.
funcshap.explainers._tree.XGBTreeModelLoader.to_integers(data:list[int]) -> np.ndarray
Handle u8 array from UBJSON.
classshap.explainers.other._maple.Maple
Simply wraps MAPLE into the common SHAP interface.
classshap.explainers.other._maple.TreeMaple
Simply tree MAPLE into the common SHAP interface.
classshap.explainers.pytree.TreeExplainer
A pure Python (slow) implementation of Tree SHAP.
funcshap.links.identity(x:npt.NDArray[Any] | float) -> npt.NDArray[Any] | float
A no-op link function.
methodshap.maskers._composite.Composite.data_transform(*args:Any) -> list[Any]
Transform the argument
methodshap.maskers._composite.Composite.mask_shapes(*args:Any) -> list[Any]
The shape of the masks we expect.
classshap.maskers._image.Image
Masks out image regions with blurring or inpainting.
methodshap.maskers._image.Image.save(out_file)
Write a Image masker to a file stream.
classshap.maskers._masker.Masker
This is the superclass of all maskers.
classshap.maskers._tabular.Tabular
A common base class for Independent and Partition.
methodshap.maskers._tabular.Tabular.save(out_file)
Write a Tabular masker to a file stream.
classshap.maskers._text.SimpleTokenizer
A basic model agnostic tokenizer.
classshap.maskers._text.Text
This masks out tokens according to the given tokenizer.
methodshap.maskers._text.Text.load(in_file, instantiate=True)
Load a Text masker from a file stream.
methodshap.maskers._text.Text.mask_shapes(s)
The shape of the masks we expect.
methodshap.maskers._text.Text.save(out_file)
Save a Text masker to a file stream.
methodshap.maskers._text.Text.shape(s)
The shape of what we return as a masker.
classshap.maskers._text.Token
A token representation used for token clustering.
classshap.models._model.Model
This is the superclass of all models.
methodshap.models._model.Model.save(out_file:BinaryIO) -> None
Save the model to the given file stream.
classshap.models._text_generation.TextGeneration
Generates target sentence/ids using a base model.
classshap.plots._decision.DecisionPlotResult
The optional return value of decision_plot.
funcshap.plots._decision.multioutput_decision(base_values, shap_values, row_index, **kwargs) -> DecisionPlotResult | None
Decision plot for multioutput models.
classshap.plots._force.AdditiveForceVisualizer
Visualizer for a single Additive Force plot.
funcshap.plots._force.save_html(out_file, plot, full_html=True)
Save html plots to an output file.
funcshap.plots._force_matplotlib.draw_additive_plot(data, figsize, show, text_rotation=0, min_perc=0.05)
Draw additive plot.
funcshap.plots._force_matplotlib.draw_bars(out_value, features, feature_type, width_separators, width_bar)
Draw the bars and separators.
funcshap.plots._force_matplotlib.format_data(data)
Format data.
funcshap.plots._monitoring.monitoring(ind, shap_values, features, feature_names=None, show=True)
Create a SHAP monitoring plot.
classshap.plots._style.StyleOptions
A TypedDict of partial updates to a style configuration
funcshap.plots._style.get_style() -> StyleConfig
Return all currently active global style configuration options.
funcshap.plots._style.load_default_style() -> StyleConfig
Load the default style configuration.
funcshap.plots._style.set_style(_style:StyleConfig | None=None, **options:Unpack[StyleOptions]) -> None
Set options in the currently active global style configuration.
funcshap.plots._text.values_min_max(values, base_values)
Used to pick our axis limits.
funcshap.plots._utils.fill_counts(partition_tree)
This updates the
funcshap.plots.colors._colorconv.xyz2rgb(xyz)
XYZ to RGB color space conversion.
classshap.utils._exceptions.ExplainerError
Generic errors related to Explainers
funcshap.utils._general.format_value(s:Any, format_str:str) -> str
Strips trailing zeros and uses a unicode minus sign.
funcshap.utils._general.ordinal_str(n:int) -> str
Converts a number to and ordinal string.
funcshap.utils._general.sample(X:_ArrayT, nsamples:int=100, random_state:int=0) -> _ArrayT
Performs sampling without replacement of the input data ``X``.
funcshap.utils._legacy.convert_to_model(val, keep_index=False)
Convert a model to a Model object.

About this data

These signatures were extracted from the public source of shap/shap using Python's ast module. Argument names, default values, type annotations and return types are taken verbatim from the code. Implementation bodies are never stored. See how it works for details.

Back to all 805 libraries