API Reference¶
This page is generated from the public objects declared in src/xwhy/**/__init__.py. Adding or removing an object from a package's __all__ updates this index during the next documentation build.
Select an object for its full signature, parameters, return values, and documented members.
Explainers¶
| Object | Description |
|---|---|
xwhy.ImageClassificationExplainer |
Explainer for image classification models. |
xwhy.ImageGenerationAndEditingExplainer |
Explainer for image generation and editing tasks. |
xwhy.LLMExplainer |
Explainer for LLM tasks integrating the full GSMILE pipeline. |
xwhy.PointCloudExplainer |
Explainer for Point Cloud classification tasks. |
xwhy.TabularExplainer |
Explainer for Tabular models utilizing the SMILE algorithm. |
xwhy.TextExplainer |
Explainer for natural language processing (NLP) text classification tasks. |
Core abstractions and results¶
| Object | Description |
|---|---|
xwhy.core.BaseExplainer |
Abstract base class for all xwhy explainers. |
xwhy.core.BaseXWhyResult |
Abstract base container for shared explanation results. |
xwhy.core.ExplainerConfig |
Explainer config. |
xwhy.core.ImageClassificationConfig |
Configuration for the Image Classification explainer. |
xwhy.core.ImageGenerationAndEditingConfig |
Configuration for the Image Generation and Editing explainer. |
xwhy.core.LLMConfig |
Configuration for the LLM explainer. |
xwhy.core.PointCloudConfig |
Configuration for the Point Cloud explainer. |
xwhy.core.TabularConfig |
Configuration for the Tabular explainer. |
xwhy.core.TextConfig |
Configuration for the Text explainer. |
xwhy.core.XWhyError |
Base exception for xwhy package. |
Plots¶
| Object | Description |
|---|---|
xwhy.plots.bar |
Create a bar plot of a set of XWhy values. |
xwhy.plots.BaseTextPlotter |
Abstract base class for text plots. |
xwhy.plots.beeswarm |
Create a beeswarm plot (requires multiple instances/2D data). |
xwhy.plots.create_clean_3d_layout |
Create a minimal 3D Plotly layout without axes clutter. |
xwhy.plots.create_point_cloud_trace |
Create a Plotly 3D scatter trace for point cloud visualization. |
xwhy.plots.create_rotation_frames |
Generate camera rotation frames for 3D Plotly animations. |
xwhy.plots.decision |
Visualize model decisions using cumulative XWhy values. |
xwhy.plots.display_plotly_figure |
Display a Plotly figure to persist after reopening a Jupyter notebook. |
xwhy.plots.embedding |
Use the XWhy values as an embedding projected to 2D (requires 2D data). |
xwhy.plots.Explanation |
Public XWhy API object Explanation. |
xwhy.plots.force |
Visualize the given XWhy values with an additive force layout. |
xwhy.plots.get_image_and_mask |
Extract the segmented image and feature mask from an explanation result. |
xwhy.plots.group_difference |
Plot the difference in mean XWhy values between two groups (2D data). |
xwhy.plots.heatmap |
Create a heatmap plot (requires multiple instances/2D data). |
xwhy.plots.image |
Plot XWhy values for image inputs. |
xwhy.plots.image_boundaries |
Plot the image superpixel boundaries for the most important features. |
xwhy.plots.image_heatmap |
Plot a heatmap of feature importance over image superpixels. |
xwhy.plots.image_regions |
Plot highlighted positive and negative regions of the explanation. |
xwhy.plots.image_regions_side_by_side |
Plot positive and combined positive/negative regions side by side. |
xwhy.plots.image_to_text |
Plot XWhy values for image inputs with text outputs. |
xwhy.plots.initjs |
Do nothing; kept so SHAP-style notebooks keep running unchanged. |
xwhy.plots.monitoring |
Create a monitoring plot over time or indices (requires 2D data). |
xwhy.plots.NativeHeatmapPlotter |
Native matplotlib implementation of text heatmap plot. |
xwhy.plots.partial_dependence |
Plot the partial dependence of a model on a single feature. |
xwhy.plots.plot_3d_mesh |
Create a 3D mesh visualization with a rotation animation. |
xwhy.plots.plot_3d_point_cloud |
Create a 3D point cloud visualization with a rotation animation. |
xwhy.plots.plot_colored_3d_point_cloud |
Create a colored 3D point cloud visualization with rotation. |
xwhy.plots.plot_dataset |
Plot dataset or single point with flexible matplotlib-style arguments. |
xwhy.plots.plot_explanation_waterfall |
Create a dynamic waterfall plot for explanation method coefficients. |
xwhy.plots.plot_feature_bar_chart |
Generate and optionally save a Plotly bar chart for feature contributions. |
xwhy.plots.plot_feature_box_plot |
Generate and optionally save a Plotly box plot for feature contributions. |
xwhy.plots.plot_feature_contributions |
Visualize feature contributions using a horizontal bar chart. |
xwhy.plots.plot_image |
Display an image (Tensor, Numpy, PIL, or Path). |
xwhy.plots.plot_method_contributions |
Visualize feature contributions for a given explanation method. |
xwhy.plots.plot_point_cloud |
Visualize explanation over point cloud by highlighting important clusters. |
xwhy.plots.plot_point_cloud_clusters |
Visualize clustered point cloud segments in 3D. |
xwhy.plots.scatter |
Create a dependence scatter plot (requires multiple instances/2D data). |
xwhy.plots.text |
Plot a text explanation using coloured, self-contained HTML. |
xwhy.plots.text_heatmap |
Plot a heatmap visualization for the given explanation result. |
xwhy.plots.TextPlotterFactory |
Factory for creating text visualization instances. |
xwhy.plots.TextPlotterType |
Enumeration for supported text plot backends. |
xwhy.plots.violin |
Create a violin plot (requires multiple instances/2D data). |
xwhy.plots.waterfall |
Plot an explanation of a single prediction as a waterfall plot. |
Distances¶
| Object | Description |
|---|---|
xwhy.distance.AndersonDarlingDistance |
Anderson-Darling distance metric (Custom Implementation). |
xwhy.distance.BaseDistance |
Abstract base class for unified distance implementations. |
xwhy.distance.BaseNumericDistance |
Base class for handling dimensionality of numerical distances. |
xwhy.distance.CosineDistance |
Cosine distance metric. |
xwhy.distance.CvMDistance |
Cramer-Von Mises distance metric (Custom Implementation). |
xwhy.distance.DistanceNormalizer |
Normalize distance values into similarity scores. |
xwhy.distance.DistanceType |
Enumeration for supported distance metrics. |
xwhy.distance.DTSDistance |
DTS distance metric (Custom Implementation: Combination of AD and CVM). |
xwhy.distance.KSDistance |
Kolmogorov-Smirnov distance metric (Custom Implementation). |
xwhy.distance.KuiperDistance |
Kuiper distance metric (Custom Implementation). |
xwhy.distance.WassersteinDistance |
Wasserstein distance metric (Custom Implementation). |
xwhy.distance.WMDDistance |
Word Mover's Distance metric for raw text strings. |
Models¶
| Object | Description |
|---|---|
xwhy.models.BaseClassification |
Base class for all classification implementations. |
xwhy.models.BaseEmbedding |
Base class for all embedding implementations. |
xwhy.models.BaseSegmentation |
Base class for all segmentation implementations. |
xwhy.models.ClassificationFactory |
Manage classification model instantiation via a registry. |
xwhy.models.ClassificationType |
Supported classification backends. |
xwhy.models.EmbeddingFactory |
Manage embedding model instantiation via a registry. |
xwhy.models.EmbeddingType |
Supported embedding backends. |
xwhy.models.SegmentationFactory |
Manage segmentation model instantiation via a registry. |
xwhy.models.SegmentationType |
Supported segmentation backends. |
xwhy.models.TabularModelAdapter |
Wrap tabular models to provide a unified prediction interface. |
xwhy.models.TorchvisionClassification |
Classification backend for standard torchvision models. |
xwhy.models.TorchvisionSegmentation |
Segmentation backend for standard torchvision models. |
xwhy.models.Word2VecEmbedding |
Word2Vec embedding backend. |
Providers¶
| Object | Description |
|---|---|
xwhy.providers.BaseProvider |
Abstract interface for external AI providers. |
xwhy.providers.OpenAIProvider |
OpenAI implementation of the provider interface. |
xwhy.providers.ProviderFactory |
Factory for provider implementations. |
xwhy.providers.ProviderResolver |
Resolver mapping provider types to default instantiation logic. |
xwhy.providers.ProviderType |
Supported provider types. |
Perturbation¶
| Object | Description |
|---|---|
xwhy.perturbation.BasePerturbation |
Abstract base class for perturbation strategies. |
xwhy.perturbation.ImagePerturbation |
Perturbation strategy for images using superpixels and Bernoulli sampling. |
xwhy.perturbation.TextPerturbation |
Generate binary perturbations for text. |
Surrogate models¶
| Object | Description |
|---|---|
xwhy.surrogate.BaseSurrogate |
Abstract base class for all surrogate models. |
xwhy.surrogate.LinearRegressionSurrogate |
Surrogate wrapper for linear models like OLS and Ridge. |
xwhy.surrogate.SurrogateFactory |
Factory class for instantiating surrogate models. |
xwhy.surrogate.SurrogateTrainer |
Service for training and evaluating surrogate models. |
xwhy.surrogate.SurrogateType |
Enumeration for supported surrogate model types. |
xwhy.surrogate.TreeBasedSurrogate |
Surrogate wrapper for tree-based models like Random Forest and XGBoost. |
Metrics¶
| Object | Description |
|---|---|
xwhy.metrics.calculate_stability_score |
Calculate stability metrics between two explanation results. |
xwhy.metrics.calculate_token_auc |
Calculate the Area Under the ROC Curve (AUC) for token importance. |
xwhy.metrics.ImageCoverageMetrics |
Metrics for evaluating image explanation coverage. |
xwhy.metrics.RegressionMetricResult |
Data container for regression evaluation metrics. |
xwhy.metrics.RegressionMetrics |
Utility for calculating comprehensive regression metrics. |
Configuration¶
| Object | Description |
|---|---|
xwhy.config.Settings |
Global application settings. |
xwhy.settings |
Public XWhy API object settings. |
Utilities¶
| Object | Description |
|---|---|
xwhy.utils.denormalize_tensor |
Reverse the normalization applied to an image tensor. |
xwhy.utils.load_image_as_tensor |
Load an image from disk and apply preprocessing transforms. |
xwhy.utils.numpy_image_to_tensor |
Preprocess a numpy image array using the specified transforms. |
xwhy.utils.tensor_to_numpy_image |
Convert a batch tensor (1, C, H, W) to a NumPy image (H, W, C). |
Datasets¶
| Object | Description |
|---|---|
xwhy.datasets.download_i2ebench_dataset |
Download the I2EBench dataset from Google Drive and extract it. |
xwhy.datasets.load_i2ebench_data |
Parse the I2EBench dataset with limits and file validation. |
Module-level documentation is also generated under reference/xwhy/ for maintainers and existing deep links, but it is intentionally excluded from the public API index.