Details

The AIX360 project is structured around a core set of Explainability Algorithms, which are central to its functionality. These algorithms rely on Data Handlers & Loaders to prepare and ingest data, ensuring that the input is in a suitable format for processing. The foundational Base Classes & Interfaces define the common contracts and abstract structures that all algorithms adhere to, promoting consistency and extensibility. To support the complex operations of the core algorithms, Algorithm Support Utilities provide essential helper functions and transformations. Once explanations are generated, the Metrics & Evaluation component is responsible for assessing their quality and effectiveness. Finally, Examples & Tutorials serve as a crucial entry point for users, demonstrating practical applications and interactions across the Data Handlers & Loaders, Explainability Algorithms (Core), and Metrics & Evaluation components, illustrating the typical data flow and usage patterns within the library. This modular design facilitates clear separation of concerns and promotes maintainability and scalability.

Data Handlers & Loaders

Manages data ingestion, format conversion, and preparation of datasets for explainability algorithms.

Base Classes & Interfaces

Defines foundational abstract classes and interfaces, establishing common contracts for explainability algorithms and related components.

Explainability Algorithms (Core)

Encapsulates the primary explainability algorithms responsible for generating explanations, fitting models, and making predictions.

Algorithm Support Utilities

Provides helper functions and classes that support the core explainability algorithms, handling data transformations, internal model interactions, and specific sub-routines.

Metrics & Evaluation

Offers functionalities for evaluating the quality and effectiveness of explanations generated by the algorithms, including various metrics and tools for quantitative assessment.

Examples & Tutorials

Contains runnable examples and tutorial code demonstrating how to use the AIX360 library and its various explainability algorithms.