Predicting the impact of size of uncertainty events on the construction cost of highway projects using ANFIS

Abstract

This study examines the use of Adaptive Neuro-Fuzzy Inference System (ANFIS) as a machine learning technique in the prediction of the impact size of uncertainty events on construction cost of highway projects and whether this technique is more accurate than the classical statistical methods. The rationale for the study stems from the availability of several techniques such as regression analysis and machine learning for developing predictive models of relationships of various variables in the construction industry. However, there has been limited research undertaken to compare the accuracy of the available techniques. The success or failure of prediction depends on the credibility of the prediction method. In this study, the predicted impact size of 76 uncertain events on the construction of highway projects using ANFIS as an intelligence machine learning method and Stepwise Regression Analysis (SRA) as a classical statistical method were compared to delineate the ability and accuracy of the ANFIS prediction technique. The comparison of calculated R-Value and two error tests for ANFIS and SRA show that the constructed ANFIS model has a higher performance than the SRA method in both fitness and reliability of the prediction model. Also, the performance comparison showed that ANFIS is a good tool for predicting the impact of uncertainty events on construction project cost. Based on these findings, the study concludes that the use of intelligent methods such as ANFIS will minimise the potential inconsistency of correlations in construction cost and time prediction. The model developed enables cost engineers to estimate the construction cost with a higher degree of accuracy.

Using Machine Learning to Predict Cost Overruns in Construction Projects

“…Machine learning methods including linear regression, support vector machines, and artificial neural networks have been used to estimate building costs in a variety of ways, including preliminary cost assessment (Jaafari, A., Pazhouhan, I., & Bettinger, P., 2021), cost contingency analysis, and risk assessment (Zhang, H., Li, H., Zhu, Y., & Fang, Y., 2019). These studies have shown that machine learning techniques can effectively capture the complex relationships between project parameters and costs, providing more accurate and reliable estimates (Alireza, M., & Abimbola, W., 2019). Despite these promising findings, the application of machine learning in construction cost estimation is still a relatively new area of research, with many studies limited by small sample sizes or narrow scopes (Nguyen Van, T., & Nguyen Quoc, T. , 2021).…”