Performance comparison between ordinal logistic regression and classification tree

In statistical applications, parametrical and nonparametrical methods are often applied to identify the relationship between the dependent and independent variables. In this thesis, among these methods the comparison of the classification performance of the classification tree model and ordinal logistic regression model is purposed. For this aim, an application study is conducted to determine the factors effecting the unit price of real estate for sale in Eskisehir. The performance comparison of the methods is presented by splitting dataset into three parts: training, validation and test using 5-fold cross validation. Based on the results, classification tree algorithm over performed the ordinal logistic regression model.