Traditional architectural styles in Northwest Henan: a machine learning analysis of influencing factors

As a significant component of traditional Chinese architectural culture, the architectural styles of northwest Henan embody profound historical, cultural, and artistic values. This study investigates the characteristics of materials, technology, decoration, and functions across four distinct architectural styles: adobe houses, stone houses, cave houses and brick, and tile houses in northwest Henan. A comprehensive system of influencing factors for architectural styles has been developed, encompassing 30 characteristic indicators categorized into five dimensions: natural factors, functional requirements, economic considerations, construction technology, and cultural beliefs. Furthermore, a questionnaire survey was conducted to assess the importance of these characteristic indicators across the four types of architectural styles. Utilizing data from this survey, this paper employs a random forest classification model to identify key features associated with various architectural styles. The SHAP value method is applied to illustrate the differential impact of these key features among different styles. The predictive results indicate that the identified important features can effectively forecast various architectural styles with an accuracy rate of 91.3%. Finally, this paper discusses the underlying logic behind these important features and their implications for preserving historical architectural relics as well as fostering continuity in architectural culture within the context of northwest Henan’s actual circumstances.