Telecommunication Customer Churn with Responsible AI: A Predictive Model Debugging and Business Decision Making

Customer churn means a customer stop doing business with telecommunication, that is not using the service of the company again after using it for a specific period. Acquiring new customer is five to six times costlier than retaining existing customers, therefore business that invest in using Artificial intelligence to reduce customer churn risk is wise. This study leverages Artificial Intelligence (AI) and causal analysis within a Responsible AI framework to assess feature influence on churn and mitigate model bias. Through interpretable machine learning, the study enhances model transparency for stakeholders. Causal inference analysis identifies key variables affecting churn. Evaluation metrics indicate LightGBM outperforms other models, with contract type, tenure, and dependents as significant predictors. This paper demonstrates the application of Responsible AI principles in addressing customer churn. Customer churn occurs when a customer stop conducting business with a telecommunications firm, which indicates that they do not use the company’s service again after a certain length of time. Acquiring new customers is five to six times more expensive than keeping existing customers, thus businesses who engage in artificial intelligence to reduce customer churn risk are savvy. This study uses AI and causal analysis within a Responsible AI framework to examine feature influence on churn and reduce model bias. The study improves model transparency for stakeholders by including interpretable machine learning. Causal inference study finds the important variables that influence churn. LightGBM beats other models in terms of evaluation measures, with contract type, tenure, and dependents serving as key predictors. The findings show that month-to-month contracts and greater total expenses increase churn, while longer contracts and specific internet services decrease it. This paper highlights the use of Responsible AI concepts to solve consumer attrition.