What is it about?
Construction delays cost millions. But why do project timelines keep missing the mark? Using data from 111 real government projects, this study developed an AI-powered formula that accurately predicts how long a construction project will take. It can be used by engineers and planners with just a basic calculator.
Featured Image
Photo by Synth Mind on Unsplash
Why is it important?
This study presents a timely solution to persistent construction delays by developing a data-driven ANN model that predicts government project durations with high accuracy. Unlike typical black-box models, it produces the first-ever simple, single-line equation usable with a basic calculator. Trained on 111 real projects and supported by ANOVA and Shapley analyses, the model highlights key factors like cost, locality, and relativity. Its practical and accessible nature empowers engineers and project managers to make fast, reliable estimates, bridging the gap between AI innovation and everyday construction planning.
Perspectives
Writing this article was particularly meaningful to me because it bridges my interests in data science and practical construction management. The idea of transforming a complex neural network into a single-line, calculator-friendly equation came from years of seeing engineers struggle with inaccessible tools. I hope this work empowers practitioners, especially those in public sector projects, to make more informed, confident decisions. More than just numbers and predictions, it’s about improving planning transparency and minimizing costly delays. If this model helps even one project run more smoothly, I’ll consider that a success.
Ahmad Kueh
Universiti Malaysia Sarawak
Read the Original
This page is a summary of: Data-driven artificial neural network formulated multi-factored expression for predicting construction duration in government projects, Engineering Construction & Architectural Management, June 2025, Emerald,
DOI: 10.1108/ecam-10-2024-1426.
You can read the full text:
Contributors
The following have contributed to this page
