Comparative Analysis of Machine Learning Models in Inventory Stock Prediction

Abstract

Accurate inventory forecasting plays an important role in helping organizations maintain sufficient stock while avoiding unnecessary storage and shortage-related costs. With changing customer demand and increasing business complexity, conventional forecasting methods may not always provide reliable predictions. This study investigates the application of machine learning techniques for predicting inventory stock levels using historical inventory data. Three machine learning algorithms—Decision Tree, Random Forest, and Gradient Boosting—are implemented and comparatively evaluated. Their predictive performance is assessed using Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and R² Score. The comparison focuses on determining how effectively each model can capture patterns within inventory data and estimate future stock requirements. The results indicate differences in predictive performance among the models, with Gradient Boosting achieving the strongest overall performance in the conducted analysis. The study demonstrates that machine learning can serve as an effective decision-support approach for inventory planning and can help organizations make more informed stocking decisions, reduce the likelihood of excess or insufficient inventory, and improve overall inventory management.