“An explainable machine learning decision support system for strategic IT outsourcing with evidence from an emerging market” (2026), by Yaser Sobhanifard et al. Business and Management research paper. Original abstract below, unchanged. A free PDF is available.

An explainable machine learning decision support system for strategic IT outsourcing with evidence from an emerging market

Yaser Sobhanifard, Salah Mahdi Abaadi Al-Jubouri · October 25, 2026 · Business and Management

Abstract

This study develops an explainable machine learning-based Decision Support System (DSS) to predict strategic IT outsourcing success using only pre-contractual information. Using historical data from 336 IT projects in Iraq (2020–2025), we trained four models (Logistic Regression, Decision Tree, Random Forest, XGBoost) on features such as budget, duration, vendor type, contract model, and industry. Random Forest achieved the best test-set F1-score (0.461), while XGBoost showed the strongest cross-validation performance (0.650). All machine learning models outperformed the Majority-Class Baseline in weighted F1-score. Feature importance and SHAP analysis revealed that vendor type (offshore global) and contract model (fixed-price) are the most influential success contributors—more important than raw budget. High Success projects were characterized by fixed-price contracts (75%), offshore global vendors (40%), and moderate budget intensity ($11,169/month), lower than Medium Success but higher than Low Success. SHAP also provided clear, instance-level explanations (e.g., a High Success prediction with 85.5% probability). The model could not detect Low Success projects, suggesting that failure depends on unobservable post-contractual factors. Overall, the DSS offers a practical, transparent tool for improving vendor selection in emerging markets.

Free PDF DOI: 10.1007/s44163-026-02608-y