Predicting House Prices in Turkey by Using Machine Learning Algorithms

Housing prices are an important reflection of the economy, and housing price ranges are of great interest for both buyers and sellers. The goal of this paper is to empirically conduct the best machine learning regression model for Turkish Housing Market by comparing accuracy scores and absolute deviations of test results by using Python programming language and Keras library for the five-year period from January 2015 to December 2019. This study consists of 15 explanatory variables describing (almost) every aspect of houses in Istanbul, Izmir and Ankara. These fifteen explanatory building and dwelling variables are used for each prediction model. In this study, three different data models are created by using support vector machine, feedforward neural networks and generalized regression neural networks algorithms. The experiments demonstrate that the Feedforward Neural Network model, based on accuracy, consistently outperforms the other models in the performance of housing price prediction. According to another result of the study, the most important variables in the model are the location of the house and the size of the house, while the size of the terrace is determined as the least important variable. JEL classification numbers: R10, R15, R19Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Keywords: Housing market, Zingat.com, Machine learning, House price prediction, Python programming language, Keras library.