Unsupervised machine learning for identifying key risk factors contributing to construction delays

Abstract

The present study uses unsupervised machine learning capabilities with an emphasis on K-means clustering for addressing the problem of construction delays. The primary objective is to investigate the critical risk factors that contribute to such delays, thereby enabling more efficient risk-management strategies. The study employs a large dataset compiled from contracting firms operating in developing regions. This information is a vital resource for identifying crucial risk variables. These variables are analysed and categorised using the Likert scale into five levels based on their potential influence. This systematic approach permits the development of a comprehensive understanding of the relevant factors. These risk factors are grouped to enhance comprehension of the intricate risk landscape using K-means clustering. This allows for a broader, more comprehensive understanding of the factors contributing to construction delays. The application of K-means clustering demonstrates the potential of machine learning techniques to improve conventional approaches to risk management. This empirical investigation significantly expands the existing body of construction risk-management knowledge. It offers invaluable insights into various project stakeholders, allowing for more informed decision-making. Notably, the clustering analysis results provide a practical, user-friendly tool. This tool can assist project managers in enhancing their risk foresight, drafting more effective plans and developing robust mitigation strategies. Consequently, this research offers the potential for substantial improvements in project timeline adherence, thereby substantially reducing the impact of construction delays in developing nations.