Government Construction Project Budget Prediction Using Machine Learning
K-Nearest Neighbors (KNN), an ML algorithm, was used to classify over-budget projects and demonstrated that it can be used to predict the over- budget construction projects for the Thai government.
Abstract
The construction industry could not avoid the technology disruptive era. Therefore, the Thai government has created a new policy and directed all departments to implement big data technology. Big data technology includes Machine Learning (ML). The present study attempts to predict over-budget construction projects using an ML algorithm. Data were collected from the comptroller general’s department of Thailand for over-budget project cases. Information about 692 projects completed in Thailand in 2019, covering all types of construction projects, was collected and analyzed. ML, an analytical technique for big data technology, was used as a tool in this study. In addition, k-Nearest Neighbors (KNN), an ML algorithm, was used to classify over-budget projects. The input data have four attributes: department of project, construction site location, type of project, and methods of procurement; the output is a yes/no decision on whether a project has been over budget. The dataset was preprocessed for analysis and modeled using the KNN function in Python 3. According to the test results, the KNN model achieves an accuracy (precision) of 0.86. Finally, the developed model has demonstrated that it can be used to predict the over-budget construction projects for the Thai government.