Real Estate Assessment of Agricultural Lands Outside the Zoning Plan with Artificial Neural Networks and Multiple Regression Analysis Methods: The Case of Aksaray, Bahçesaray and Kırımlı Rural Districts

Abstract

The aim of this study is to determine the criteria affecting the market value of agricultural lands in the rural Bahçesaray and Kırımlı districts, which are outside the zoning plan in the Aksaray Province Central District, where agricultural production continues, and to provide a value estimation and value map with the help of mass valuation methods and geographic information systems (GIS). Using the sales data from 125 parcels in the study area, the market value of the immovables for which the value is unknown in the region, was estimated. The most frequently used criteria in the assessment of agricultural lands were determined, and the valuation was carried out with Multiple Regression Analysis (MRA) and Artificial Neural Networks (ANN). By means of the assessment and the valuation study, the performance of the valuation methods was compared, and it was determined that the best result according to the test data was the valuation with ANN. In the performance analysis conducted with ANN, the Coefficient of Determination (R²) =0.87, Mean Absolute Percentage Error (MAPE)=0.192, Mean Square Error (MSE)=0.047 and Root Mean Square Error (RMSE)=0.059 was found. Moreover, according to the proportional standards guide determined by the International Association of Assessing Officers (IAAO), the performance measurement, the values derived for the Coefficient of Dispersion as (COD)=19.58 and Price-Related Differential as (PRD)=1.02 were also found to be within acceptable limits. Value maps were generated with the results obtained from the calculation. These maps will be drawn in the region for purposes of purchase and sale, expropriation, credit facility, and taxation, etc. And thus, the maps can be used as a base for the aforementioned subjects.