An Exploratory Machine Learning Analysis of Infrastructure Approval Drivers in Nigeria

Infrastructure projects are important to Nigeria's socioeconomic development. However, the basis for selecting, implementing, and commissioning such projects often remains unclear. Using a limited dataset of 30 projects in Nigeria between 2019 and 2024, this study employs expert-rated evaluations and machine learning to explore the underlying drivers of infrastructure approval. Projects were evaluated across categories, including economic impact, social value, safety, environment, technological advancement, and political biases. A Random Forest Model trained on expert ratings achieved 67% accuracy, with economic impact and safety enhancement emerging as the most influential decision factors. The analysis revealed critical approval thresholds, where projects scoring below moderate influence (3.0) on economic impact had less than a 45% likelihood of approval. Notably, while political bias received low expert ratings, it significantly reduced approval probabilities when present. The study introduces practical innovations for systematically comparing expert assessments with data driven driver weights and an interactive tool for simulating approval scenarios. The research contributes the first Machine Learning analysis of Nigeria’s infrastructure approval drivers, offering actionable insights for optimizing project selection. The methodology demonstrates how machine learning can augment expert judgment in public investment decisions, particularly in resource constrained nations.