Özet:

In addition to accommodation, housing, which is an investment tool, has to be evaluated over the market value, which depends on many factors in the marketing and sales stages, since it does not have an absolute value and is completely relative. With this study, an Artificial Neural Network (ANN) model, which is one of the artificial intelligence methods, has been developed in order to determine the current values of the houses quickly and accurately. For this purpose, a total of 149 houses for sale, which were announced on an e-commerce site in Turkey, located in different parts of Ankara's Kecioren district, were taken into consideration to create NNS models. Artificial neural network models were created by digitizing 11 parameters that are effective in determining the current value of a house. By changing the hidden layer, number of neurons and activation function, 20 models were established and the most suitable hidden layer, number of neurons and activation function were determined. It has been seen that with the mean square error (MSE) 0.000197, the regression (R) %94,31 and the accuracy rate %91,59, this chosen ANN architecture gave successful results in determining the fair values of the houses.

Anahtar Kelimeler: