SELECTING A STANDARD SET OF ATTRIBUTES FOR THE DEVELOPMENT OF MACHINE LEARNING MODELS OF BUILDING PROJECT COST ESTIMATION

Abstract

Accurate cost estimation is a critical aspect of successful construction projects, and the application of machine learning offers promising advancements in this domain. However, to achieve reliable cost predictions, the selection of a standardized set of attributes that significantly influence model performance is essential. This research addresses the research gap by investigating the systematic clarification of a standard set of attributes for machine learning models in building cost estimation. Firstly, plenty of attributes were summarized by literature review, then by questionnaire surveying and focus group discussion of the Delphi study period, the final 68 ranked attributes were determined and formulated the attribute set of building data. The findings of this research are beneficial to improve the accuracy of estimation by providing the essence of developing a building cost estimation of machine learning because the domain researcher can refer to these listed attributes to determine the lay structure of a new model.