Logistics operations in panelized construction are vital daily tasks that connect the panel manufacturing facility to the job site. Although logistics operations are both important and prevalent in panelized construction, the cost of logistics has yet to be fully understood by either industry or academia due to the complicated relationship between multiple factors in logistics demands and operations. In practice, logistics is considered as an overhead cost that consists of various indirect or fixed costs in the panelized construction operation. As a result, logistics cost estimates are rendered inaccurate when subjected to project changes. Considering the number of construction projects over the course of a year, inaccurate logistics cost estimates are significant. Previous studies have shown that a machine learning approach could be used to predict costs that are influenced by multiple factors. To fill knowledge gaps in both research and practice, in this study machine learning based on historical logistics data is used to accurately predict logistics costs for a given project. The results from this study indicate that machine learning can be a reliable tool to predict logistics costs.
An empirical analysis of data preprocessing for machine learning-based software cost estimation
