Comparing negative binomial regression and quantile regression for modelling of housing sales

Abstract

Housing demand arises under the influence of social, cultural, economic and demographic characteristics of each country Housing requirement refers to not only a numerical value, but also a structure that must carry healthy conditions with its environment. Nowadays, both housing sales and the price of properties are in increasing trend. In this study, the factors affecting the housing sales of the Turkey’s 81 provinces for 2019 were examined. Among alternative regression methods, negative binomial regression and quantile regression models were used and analysis were made on the basis of various quantile slices. As a result of the negative binomial regression model, it was found that the crude marriage rate, divorce rate, gross domestic product, unemployment rate and crude birth rate variables had a significant effect on the model. In quantile regression model analysis, quantile slices were evaluated separately. By using CICOMP type information criteria, it was seen that negative binomial regression model gave more effective results than quantile regression model.