Using Artificial Intelligence and Machine Learning Techniques to Predict Engineering Project Performance in Terms of Duration and Cost: An Applied Study

Abstract

This study aims to assess the impact of Artificial Intelligence (AI) and Machine Learning (ML) techniques on
predicting the performance of engineering projects, focusing on project duration and cost for the maintenance
of Al-Shat Road – Tripoli, Libya. The study’s significance stems from the challenges faced by the road sector
in Libya, such as recurring delays, cost overruns, and the limited accuracy of traditional planning methods
based on engineers’ experience and the Critical Path Method (CPM).

The study employed a field-based applied methodology, collecting field data and official project reports, along
with expert evaluations from supervisors and engineers. The study sample included actual project data from
2024–2025 and 25 engineers and supervisors involved in project execution. Data were analyzed using ML
algorithms Random Forest, XGBoost, and Artificial Neural Networks (ANN), with model accuracy evaluated
using performance indicators RMSE, MAE, and R².
The results showed an 11% deviation in project duration and a 12% deviation in cost compared to planned
values, highlighting the limitations of traditional methods. Smart models demonstrated a significant
improvement in prediction accuracy, with XGBoost achieving the best performance for both duration and cost
estimation. The study also revealed that weather conditions and labor availability were the most influential
factors on duration, while material price fluctuations and design changes had the greatest impact on cost.
Based on these findings, the study recommends: adopting AI models for planning engineering projects,
monitoring performance-influencing factors, training engineers on smart tools, and establishing a centralized
project database to facilitate future application of predictive models. Additionally, it suggests expanding the
research to other infrastructure projects and utilizing Deep Learning techniques for handling large and
complex project data.
The study confirms that employing AI and ML techniques significantly improves the accuracy of duration and
cost predictions, reduces deviations, and enhances decision-making efficiency in engineering projects,
thereby improving institutional performance and minimizing the risk of project failure.