Optimising Construction Contract Rates with Z-Score and Machine Learning Approaches

This study assesses two prevalent techniques for contract rate rationalization in construction by employing Z-score statistical screening and machine learning (ML) models on a real-world tender dataset (single project; 10 tenderers). We benchmark a human-in-the-loop, two-gate workflow where Z-scores serve as an auditable initial filter for identifying anomalous bids, and ML (regression and clustering) creates context-aware price benchmarks from multivariable features. Empirically, the ML approach achieved a 38% reduction in RMSE compared to a Z-score-only baseline, while the hybrid configuration reduced spurious flags and produced tighter, more coherent pricing bands for decision support. Z-score screening retained operational value by standardizing anomaly detection and reinforcing procedural transparency, whereas ML contributed predictive sharpness and adaptability to item, project, and market heterogeneity. The findings demonstrate that combining explainable statistical rules with data-driven modeling yields measurable gains in accuracy, efficiency, and fairness for tender evaluation. The contributions of this study are threefold: (i) a head-to-head comparison of Z-score versus ML at the Bill-of-Quantities level; (ii) a governance-ready hybrid protocol that integrates auditable thresholds with model-based benchmarks; and (iii) actionable guidance for agencies seeking to institutionalize analytics within e-procurement. Although the scope is limited to a single-project dataset, the results motivate multi-project validation and real-time deployment with drift monitoring in future studies.