Developing an Integrative Data Intelligence Model for Construction Cost Estimation

Abstract

Construction cost estimation is one of the essential processes in construction management. Project cost is a complex engineering problem due to various factors affecting the construction industry. Accurate cost estimation is important in construction management and significantly impacts project performance. Artificial intelligence (AI) models have been effectively implemented in construction management studies in recent years owing to their capability to deal with complex problems. In this research, extreme gradient boosting is developed as an advanced input selector algorithm and coupled with three AI models, including random forest (RF), artificial neural network (ANN), and support vector machine (SVM) for cost estimation. Datasets were gathered based on a survey conducted on 90 building projects in Iraq. Statistical indicators and graphical methods were used to evaluate the developed models. Several input predictors were used, and XGBoost highlighted inflation as the most crucial parameter. The results indicated that the best prediction was attained by XGBoost-RF using six input parameters, with r-squared and the mean absolute percentage error equal to 0.87 and 0.25, respectively. The comparison results revealed that all AI models showed good prediction performance when applied to datasets affected by more than two parameters. The outcomes of this research revealed an optimistic strategy that can help decision makers select the influencing parameters in the early phases of project management. Also, developing a prediction model with high precision results can assist the project’s estimators in decreasing the errors in the cost estimation process.

Big Data, Data Science, and Artificial Intelligence for Project Management in the Architecture, Engineering, and Construction Industry: A Systematic Review

“…The results are contrasted with what was shown by previous reviews, such as references [36,37,84], showing that there is a growing trend of research on the subject. This review provides the classification of the articles found by PMI knowledge areas (Project Management Institute 2017), finding that the greatest interest of the academic community is in cost management [53,58,64,74,79], quality management [12,73,77], time management [57,60,62,82], and, finally, scope management [11,81,87]. There is a knowledge gap in the use of emerging technologies (big data, data science, and artificial intelligence) in the areas of stakeholder management, communications, and human resources, at least in the construction and civil works sector.…”

“…Project cost management (22 records) mainly features the use of emerging technologies to perform budget forecasting and projection [77,79,85], as well as the early detection of anomalies in financial and operational data [8,53,58]. Data science is also used to simulate different scenarios and assess the impact of decisions on the final project budget [64,68,74].…”

Using Machine Learning to Predict Cost Overruns in Construction Projects

“…Construction cost estimation is a critical component of project management, as it influences decision-making, budget allocation, and project success (Zainab, H. A., Abbas, M. B., Murizah, K., & Zainab, A.K., 2022). Several challenges commonly impact the accuracy of cost estimation, including incomplete information, uncertainties, and changing requirements (Aftab, H. M., Ismail, A. R.,Mohd, R. A., Asmi, A. A.,, 2014).…”

Construction Project Cost Prediction Method Based on Improved BiLSTM

“…However, the ARIMA model is unsuitable for capturing the nonlinearities of time series in engineering costs [2], which negatively affects prediction accuracy. To solve this problem, the support vector machine (SVM) [3][4][5], backpropagation (BP) neural network [6][7][8], and other machine learning models have been applied to cost prediction. Although such methods can effectively handle nonlinear problems, the SVM has limitations regarding data correlation processing and slow processing speed, and the BP neural network can quickly lose time series data and fall into local minimal values.…”