Leveraging Predictive Analytics for Customer Churn: A Cross-Industry Approach in the US Market

Abstract

Customer churn prediction is an important aspect of businesses to ensure their profitability in the USA. After a customer attrition calculation, which constitutes the percentage of lost customers compared to the total number of customers over a given period, companies in the USA need to develop predictive models that will help them make appropriate moves to retain customers and maximize profits. The dataset used contained highly elaborate information on customer demographics, service usage, and several indicators that are essential for the analysis of customer retention and churn. Data anonymization and protection were also considered to ensure privacy and protect sensitive company information. In this research, we develop five main machine learning models to predict customer churn using customer data from company databases and systems. The four machine learning models employed in this research include XGBoost, Random Forest, MLP(multi-layer perceptron), and Logistic Regression. The study also assesses model performance using metrics such as mean absolute error (MAE), mean squared error (MSE), and R² score.