| Title |
Study on the Development and Comparison of ANN and MLR Models for Construction Budget Estimation - Focusing on the Construction Budget for Public Building Projects - |
| Authors |
Park, Jee Young ; Kim, Han Soo |
| DOI |
https://dx.doi.org/10.6106/KJCEM.2025.26.1.031 |
| Keywords |
Construction Cost Estimation; Cost Planning; Machine Learning; Artificial Neural Network; Multiple Linear Regression |
| Abstract |
The construction budget is crucial for determining a project's feasibility, making systematic budget prediction essential for project owners. To ensure compliance with the initial budget during the design phase, it is necessary to manage the budget with a breakdown by work type. Traditional methods rely on estimators' experience and judgment using historical data, which can introduce uncertainty. The objective of the study is to develop and compare budget prediction models using Artificial Neural Network (ANN) and Multiple Linear Regression (MLR) techniques, focusing on public office building construction costs. The study evaluates the effectiveness of these models for budget prediction and identifies key characteristics. The results show that the models using ANN and MLR have error ranges of -36.3% to 58.0% and -27.7% to 32.8%, respectively, demonstrating their utility. Additionally, the MLR model exhibited more stable predictive performance for homogeneous data compared to the ANN model. |