Construction Cost Prediction Using Deep Learning with BIM Properties in the Schematic Design Phase
D.K. Park, Seok-Heon Yun
Abstract
In the planning and design stage, it is difficult to accurately predict construction costs only by estimating approximate cost. It is also very difficult to predict the change in construction costs whenever the design changes. However, using the BIM model’s attribute information and machine learning techniques, accurate construction costs can be predicted faster than when using the existing approximate cost estimate. In this study, building information such as ‘total area’, ‘floor water’, ‘usage’, and BIM attribute information such as ‘wall area’, ‘wall water’, and ‘floor circumference’ were used together to predict construction costs in the schema design stage. As a result of applying the machine learning technique using both the building design information and the BIM model attribute information, it was found that the construction cost was improved compared to the result of individual predictions of the building information or BIM attribute information. While accurately predicting construction costs using BIM’s attribute information has its limits, it is expected to provide more accuracy compared to predicting costs solely based on construction cost influencing factors.