Customer churn prediction is a key application of predictive analytics in customer relationship management, enabling organisations to identify customers at risk of leaving and implement retention strategies. This paper presents an end-to-end operational analytics framework for customer churn prediction and interactive decision support using the publicly available IBM Telco Customer Churn dataset containing 7,043 customer records. The proposed framework integrates data preprocessing, feature engineering, predictive modelling, SHAP-based interpretability, and an interactive Streamlit dashboard within a unified and reproducible machine learning pipeline. Four machine learning models – Logistic Regression, Random Forest, XGBoost, and a Multi-Layer Perceptron (MLP) Neural Network – were evaluated using an 80/20 train-test split and standard classification metrics. Logistic Regression achieved the best overall performance with an accuracy of 0.812 and an AUC of 0.863, while the MLP Neural Network achieved a competitive AUC of 0.860. The SHAP research indicated that the main churn predictors were tenure, contract type, dependents, monthly charges and total charges. The deployed dashboard enables real-time churn prediction, probability-based risk assessment, and interactive decision support for customer retention planning. The framework demonstrates how predictive analytics models can be operationalised into interpretable and deployable decision-support systems for practical business applications.
Olasomi Labo-Popoola (Fri,) studied this question.