Beyond Benchmark Accuracy: A Systematic Review and Reproducibility Audit of Machine-Learning Traffic Forecasting

A systematic methodological review, structured evidence synthesis, and reproducibility and external-validity audit of machine-learning road-traffic forecasting. The project evaluates validation design, data-leakage safeguards, baseline adequacy, reporting transparency, artifact availability, computational reproducibility, calibration, robustness, external validation, and claim–evidence congruence. Conducted by Malek Altarawneh, Independent Transportation Researcher, Jordan.