Skip to content
Open access

Explainable Artificial Intelligence and Machine Learning for Predictive Decision Support in Complex Information Systems

Nov 2026 · Computer and Decision Making: An International Journal · 0 citations

Abstract

The integration of explainable artificial intelligence (XAI) and machine learning into complex information systems requires predictive accuracy, transparency, reliability, and operational effectiveness. This study developed and evaluated an explainable machine learning framework for predictive decision support using a phishing website dataset comprising 11,055 observations and 30 predictive features. Five classifiers Logistic Regression, Random Forest, Support Vector Machine, XGBoost, and LightGBM were evaluated through group-aware cross-validation and independent testing. Model interpretability was examined using SHAP and LIME, while probability calibration, bootstrap uncertainty, and cost-sensitive threshold optimization assessed decision-support reliability. LightGBM achieved the highest ROC-AUC (0.9954), whereas XGBoost demonstrated superior classification accuracy (96.35%) and F1-score (0.9587). SHAP analysis identified URL anchor characteristics and SSL certificate status as influential predictors, with strong cross-model explanation agreement (Spearman's ρ = 0.9711). XGBoost achieved the lowest Brier score (0.02713). Under a 10:1 false-negative-to-false-positive cost scenario, optimized thresholds reduced weighted misclassification costs by 70.22% for LightGBM and 56.66% for XGBoost. These findings demonstrate that integrating explainability, calibration, and cost-sensitive evaluation can strengthen predictive decision support. External validation and operational assessment remain necessary to establish real-world generalizability.  

Read PDF

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.