Exploring the Quality-of-life Satisfaction in the Historical Fabrics of Iran Through Machine Learning Models

Abstract

Historical city centres are quite often left out of the urban development process, especially when population growth is very rapid. Therefore, they are confronted with severe difficulties affecting the quality of life (QOL) of their residents. An adequate QOL is essential to anchor the local population in these valuable historical areas. Keeping their traditional ways of life is critical to preserve their heritage, but almost no comprehensive study has been done on the subject in Iran. To address this deficiency, a multivariable analysis was carried out based on an extensive survey that counted with the participation of more than 1800 inhabitants of the old city centres of Yazd, Ardakan, Naeen and Kashan. The QOL (dependent variable) was related to 21 independent variables, covering a wide range of physical, social, economic, environmental, and institutional aspects, selected from a thorough review of the theoretical literature. To discover the patterns underlying the collected data, several different parametric and non-parametric algorithms such as CHAID, Logistic Regression, NEURAL NET, C5.0 and C&R Tree have been examined. The C5.0 model showed the highest overall accuracy and was used to select the best predictors of QOL satisfaction for the residents of these city areas: 1) quality of buildings and streets, 2) safety and security, 3) administrative services and 4) vehicle accessibility. The knowledge gathered should assist Iranian decision-makers and planners develop comprehensive regeneration plans for historical city areas and better incorporate social sustainability aspects.