The Prediction of Low-Rise Building Construction Cost Estimation Using Extreme Learning Machine

Abstract

This study aims to predict the possibility of low-rise building construction costs by applying machine learning models, and the performance of each model is evaluated and compared with ensemble methods. The artificial neural network (ANN) emerges as the top-performing individual model, attaining an accuracy of 0.891, while multiple linear regression and decision trees follow closely with accuracies of 0.884 and 0.864 respectively. Ensemble methods like maximum voting ensemble (MVE) improve the accuracy beyond individual models with an impressive accuracy rate of 0.924. Meanwhile, the stacking ensemble and averaging ensemble also demonstrate competitive performance with accuracies of 0.883 and 0.871, respectively. These findings can result in more informed decision-making, which is valuable for the real estate industry.