Chapter 6: Model Interpretability Using SHAP
In the previous two chapters, we explored model-agnostic local explainability using the LIME framework to explain black-box models. We also discussed certain limitations of the LIME approach, even though it remains one of the most popular Explainable AI (XAI) algorithms. In this chapter, we will cover SHapley Additive exPlanation (SHAP), which is another popular XAI framework that can provide model-agnostic local explainability for tabular, image, and text datasets.
SHAP is based on Shapley values, which is a concept popularly used in Game Theory (https://c3.ai/glossary/data-science/shapley-values/). Although the mathematical understanding of Shapley values can be complicated, I will provide a simple, intuitive understanding of Shapley values and SHAP and focus more on the practical aspects of the framework. Similar to LIME, SHAP also has its pros and cons, which we are going to discuss in this chapter. This chapter will cover one practical...