APPROCHE HYBRIDE DATA SCIENCE-IA POUR LA REDUCTION DES ECARTS ENTRE CONCEPTION ET EXECUTION DES PROJETS D’INFRASTRUCTURES ROUTIERES : CAS DE LA RDC

Road infrastructure projects play a crucial role in the economic and territorial development of the Democratic Republic of Congo, yet their implementation is frequently affected by significant gaps between design forecasts and actual execution outcomes, particularly in terms of cost, schedule, and construction quality. These recurrent deviations reveal the limitations of traditional project management approaches, which rely on static monitoring tools and offer limited analytical capacity in complex project environments. This paper proposes a hybrid approach combining Data Science techniques and Artificial Intelligence models to analyze, anticipate, and reduce design–execution gaps in road infrastructure projects. The methodology exploits historical data from public road projects through exploratory data analysis, unsupervised clustering, and predictive modeling of observed deviations. The results identify distinct project risk profiles and demonstrate the ability of predictive models to anticipate potential cost and schedule overruns before critical execution phases. This approach strengthens decision support and improves project governance in constrained institutional contexts, while offering a transferable framework for other developing countries facing similar infrastructure challenges.