Predicting High-Value Customers in Supply Chain Management Using Machine Learning: A Comparative Analysis

Abstract

The integration of machine learning (ML) into Supply Chain Management (SCM) has revolutionized data-driven decision-making, particularly in identifying high-value customers for strategic planning and operational efficiency. This study presents a comprehensive ML pipeline applied to a curated dataset of transactional and customer-level features to predict high-value clients. We evaluate 11 supervised learning models, including Logistic Regression, Decision Trees, Random Forest, Gradient Boosting, SVM, KNN, Naive Bayes, XGBoost, LightGBM, AdaBoost, and Bagging Classifier, using performance metrics such as accuracy, confusion matrices, and ROC-AUC scores. Remarkably, most models achieved near-perfect performance, with several attaining 100% accuracy and AUC scores. To enhance interpretability, we employ SHAP (SHapley Additive exPlanations) for feature importance analysis, revealing key drivers of customer value. Additionally, unsupervised clustering and dimensionality reduction techniques provide deeper insights into customer segmentation. Our findings demonstrate that ensemble-based models (e.g., XGBoost, LightGBM) consistently outperform traditional classifiers, while SHAP analysis improves model transparency and trustworthiness. This research offers a robust predictive framework for SCM applications, enabling precise identification of high-value customer segments to optimize marketing and supply chain strategies.