A Machine Learning Approach to Predict Customer Loyalty Based on CSR Perception Metrics

Corporate Social Responsibility (CSR) has become one of the most important engines of customer loyalty. In contrast, existing solutions do not provide the ability to make forecasts and do not incorporate the fluxes of perception data. CSR perception metrics of customer loyalty generated by structured surveys and unstructured sentiment data as a framework for machine learning-based prediction. The proposed model utilized the XGBoost classification algorithm with SHAP as an explanatory tool and AnyLogic for real-time loyalty simulation. The dataset contains five fundamental CSR drivers, such as environmental responsibility, ethical governance, community engagement, employee welfare, and transparency, overlaid on the sentiment scores. The metrics of experimental performance present the proposed model with an accuracy of 89.4%, precision of 88.2%, recall of 87.5%, F1 score of 87.8%, and AUC of 0.93, outperforming baseline classifiers. The framework has a high potential for improving customer retention strategies by making CSR-conscious decisions. Conclusively, the model connects the perception analytics and business intelligence as a way of facilitating companies to fit the CSR actions with the expectations of consumers.