A Multilayer Perceptron Model for Contractual Agreement–Driven Risk Management in Building Projects: Evidence from Kisii County, Kenya

Abstract: The construction sector continues to experience persistent contractual disputes and cost overruns arising from inefficient risk allocation, weak policy enforcement, and inadequate data-driven frameworks for predicting project performance. In Kenya, nearly 30% of public building projects have been reported by the Public Procurement Regulatory Authority (PPRA) to face disputes due to delayed payments, design variations, and inadequate contract administration. The problem of risk mismanagement has been compounded by fragmented contractual structures, particularly under the Design–Bid–Build (DBB) procurement system, which separates design and execution responsibilities. To address this challenge, the present study proposes a Multilayer Perceptron (MLP) model to predict contractual risk management performance in building projects within Kisii County. Using empirical risk data comprising the Risk Significance Index (RSIS), composite risk scores, policy enforcement indices, and contractor competence, the MLP model was trained with a hybrid Output–Hidden Weight Optimization (OWO–HWO) algorithm. The model achieved stable convergence with a final Mean Squared Error (MSE) of 0.134 and an of 0.002 across 1500 epochs, indicating computational stability but limited predictive strength due to dataset constraints. The findings demonstrate that contract type and policy enforcement significantly influence the Risk Management Performance Index (RMPI), with Design–Build contracts exhibiting higher variability compared to Design–Bid–Build projects. The study concludes that hybrid neural frameworks offer a viable approach for modeling nonlinear contractual, risk relationships in developing countries. Policy recommendations include enhancing enforcement of arbitration laws, strengthening contractor competence certification, and adopting intelligent contract analytics for project monitoring. Future research should expand data sampling and integrate advanced feature selection to improve generalization. Keywords: Contractual Risk Management, Construction Project Performance, Design–Build and Design–Bid–Build Contracts, Multilayer Perceptron, Policy Enforcement in Kenya. Title: A Multilayer Perceptron Model for Contractual Agreement–Driven Risk Management in Building Projects: Evidence from Kisii County, Kenya Author: Enock Nyanumba Ayuka, Gwaya Adednego, Winja Mathew International Journal of Novel Research in Civil Structural and Earth Sciences ISSN 2394-7357 Vol. 12, Issue 3, September 2025 - December 2025 Page No: 48-62 Novelty Journals Website: www.noveltyjournals.com Published Date: 12-October-2025 DOI: https://doi.org/10.5281/zenodo.17333581 Paper Download Link (Source) https://www.noveltyjournals.com/upload/paper/A%20Multilayer%20Perceptron%20Model-12102025-4.pdf