Enhancing Intelligence Source Performance Management through Two-Stage Stochastic Programming and Machine Learning Techniques

Lucas Wekesa Wafula^*

• Strathmore University Institute of Mathematical Sciences, Nairobi County, Nairobi, Kenya

The effectiveness of intelligence operations is shaped by the reliability and performance of human intelligence (HUMINT) sources. Yet, source behaviour is often unpredictable, deceptive, or shaped by operational context, complicating resource allocation and tasking decisions. This study presents a hybrid framework combining Machine Learning (ML) and Two-Stage Stochastic Programming (TSSP) to improve performance management under uncertainty. A synthetic dataset reflecting HUMINT operational patterns was used to train XGBoost and Support Vector Machines (SVM) models for behavioural classification and prediction of reliability and deception scores. The classifiers achieved 98\% overall accuracy, with XGBoost yielding higher precision and SVM achieving superior recall for rare but operationally significant categories. Regression models reached $R^2$ scores of 93\% for reliability and 81\% for deception. These predictive outputs were transformed into scenario probabilities for integration into the TSSP model, optimising task allocation under varying behavioural risks. Against a deterministic optimisation baseline, the hybrid framework delivered a 16.8\% reduction in expected tasking costs and a 19.3\% improvement in mission success rate. The results indicate clear advantages for scenario-based probabilistic planning over static heuristics in managing uncertain HUMINT conditions. However, since findings are based on simulated data, field validation is required before operational deployment.