Explainable Artificial Intelligence for Analytical Customer Relationship Management in Banking and Finance

We must understand the reasoning behind the decisions made by machine learning (ML) models to have better acceptability in finance domains. We employed Shapley Additive explanation (SHAP) and Local Interpretable Model-agnostic Explanation (LIME) for obtaining an explanation in solving Analytical Customer Relationship Management (ACRM) problems in banking and insurance using AI/ML techniques. Problems such as prediction of customer churn, detection of credit card and insurance fraud, and prediction of loan default are solved. Our approach employs Explainable Artificial Intelligence (XAI) within ACRM, representing a pivotal shift toward trust-building, compliance assurance, and optimization of predictive analytics. The provided explanations employ bubble, stacked bar plot, and violin plots for local and global explanations, respectively. The top 5 features are identified using LIME, while SHAP utilizes the top 10 features for comprehensive explanations.