A BIM and machine learning integration framework for automated property valuation

Abstract

Property valuation contributes significantly to market economic activities, while it has been continuously questioned on its low transparency, inaccuracy and inefficiency. With Big Data applications in real estate domain growing fast, computer-aided valuation systems such as AI-enhanced automated valuation models (AVMs) have the potential to address these issues. While a plethora of research has focused on improving predictive performance of AVMs, little effort has been made on information requirements for valuation models. As the amount of data in BIM is rising exponentially, the value-relevant design information has not been widely utilized for property valuation. This paper presents a system that leverages a holistic data interpretation, improves information exchange between AEC projects and property valuation, and automates specific workflows for property valuation. A mixed research method was adopted combining the archival literature research, qualitative and quantitative data analysis. A BIM and Machine learning (ML) integration framework for automated property valuation was proposed which contains a fundamental database interpretation, an IFC-based information extraction and an automated valuation model based on genetic algorithm optimized machine learning (GA-GBR). The main findings indicated: (1) Partial information requirements can be extracted from BIM models, (2) Property valuation can be performed in a more accurate and efficient way. This research contributes to managing information exchange between AEC projects and property valuation and supporting automated property valuation. It was suggested that the infusion of BIM, ML and other emerging digital technologies might add values to property valuation and the construction industry.