Investigation of Housing Sales with Artificial Neural Networks Method: The Case of Ankara Province

Housing is one of the important sectors of Turkey economy. Housing is a heterogeneous commodity and basic shelter. There are various factors that affect housing industry and housing sales in the economy. The housing sales figures, exchange rate of dollar, consumer confidence index, industrial production index, construction production index are among the factors influencing housing industry and housing sales data. Ankara, the capital and the second largest city of Turkey, has different housing sales and housing prices. Artificial neural networks which were inspired by the functioning of the human brain have been used for taking decision and prediction. This study aims to predict housing sale prices of Ankara by taking into consideration such factors as exchange rate of dollar, rate of interest, consumer confidence index, industrial production index and construction confidence index of 82 months between 2013-2019 years (from 2013 January to 2019 November) with artificial neural network system. The dependent variable is housing sale data of previous years while the independent variables are exchange rate of dollar, rate of interest, consumer confidence index, industrial production index and construction confidence indexes. As a result of these predictions, the error mean square of the feedforward neural network is estimated at r2=0,99. The error mean square of the Elman network is estimated at r2=0,98. It is determined that housing sale data by the feedforward neural network provide a more accurate result than that of Elman network and the feedforward neural network is stated as more useful.