ORHAN DOĞAN
Nassirou BANDE
Yunus GENÇ
Fatih Çağatay AKYÖN
Ramazan TANÇ
Abstract
With the rapid development in the construction sector in recent years, housing sales, which is one of the economic investments, have accelerated, making an objective valuation difficult and making it impossible to predict the real price due to the fact that there are too many parameters in the valuation stage and there is no definite formula. In addition, in real estate appraisal processes, the use of artificial neural networks (ANN), which is one of the artificial intelligence methods, has made it attractive to perform and adapt machine learning using examples, to provide information about unprecedented examples, to work fast and easy to identify, to provides solutions to complex problems, to work with little information. In this study, ANN models were created by rearranging the quantification values in an existing study by creating ANN models with 14 parameters that are effective in determining the fair value of a total of 220 houses for sale advertised on an e-commerce site in different neighborhoods of Yenimahalle district of Ankara/Turkey, and it was observed that with the use of rearranged quantification values, the ANN architecture selected with mean square error (MSE) 0.000016, regression (R) 95.99% and accuracy rate 91.73% gave more successful results in predicting the house price.
Keywords
Artificial Neural Networks, ANN, Quantification Values, Housing Valuation, Fair Value.
