Abstract

This study investigates the application of machine learning to predict the success of Agile training sales outcomes, aiming to enhance decision-making and performance management in Agile organizations. Using historical sales and monthly performance data, two predictive algorithms Random Forest and XGBoost were developed and evaluated to classify transactions based on their likelihood of success, represented by the “Won” status. The models incorporated key performance indicators such as Target Achievement Ratio, Sales Target Difference, and Sales Value, combined with categorical attributes like course type and sales category. The results show that both models performed effectively, with XGBoost achieving higher accuracy (87.9 percent) and recall (0.97), demonstrating its superior capability in identifying successful transactions. To enhance interpretability, SHAP (SHapley Additive exPlanations) analysis was used to determine the most influential features affecting predictions. The findings reveal that goal-related performance metrics, particularly Target Achievement Ratio and Sales Target Difference, were the strongest determinants of success, while categorical factors contributed less significantly. These results indicate that consistent goal attainment and effective performance monitoring are central to sales success in Agile training environments. The study contributes to Agile management literature by demonstrating how predictive and explainable artificial intelligence can improve forecasting accuracy, support data-driven decision-making, and promote continuous improvement within Agile business frameworks.

Keywords: Agile Management, Machine Learning, Predictive Analytics, Sales Forecasting, Explainable Artificial Intelligence (XAI)

How to Cite:

Alzahrani, K. & Alzahrani, A. A., (2026) “A Comparative Study of XGBoost and Random Forest for Predicting Agile Training Sales Success Using Explainable Machine Learning Models”, Agile Management 1(3), 212-225. doi: https://doi.org/10.63913/am.v1i3.137