Summary
In this chapter, you learned about the various model explainability methods used to explain black-box models. Some of these are model-agnostic, while some are model specific. Some of these methods provide global interpretability, while some of them provide local interpretability. For most of these methods, visualizations through plots, graphs, and transformation maps are used to qualitatively inspect the data or the model outcomes; while for some of the methods, certain examples are used to provide explanations. Statistics and numerical metrics can also play an important role in providing quantitative explanations.
In the next chapter, we will discuss the very important concept of data-centric XAI and gain a conceptual understanding of how data-centric approaches can be leveraged in model explainability.