The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Download
Publications Copernicus
Download
Citation
Articles | Volume XLVI-4/W5-2021
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLVI-4/W5-2021, 21–26, 2021
https://doi.org/10.5194/isprs-archives-XLVI-4-W5-2021-21-2021
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLVI-4/W5-2021, 21–26, 2021
https://doi.org/10.5194/isprs-archives-XLVI-4-W5-2021-21-2021

  23 Dec 2021

23 Dec 2021

USING SVR AND MRA METHODS FOR REAL ESTATE VALUATION IN THE SMART CITIES

A. U. Akar and S. Yalpir A. U. Akar and S. Yalpir
  • Department of Geomatics Engineering, Konya Technical University, 42100, Konya, Turkey

Keywords: Real Estate Valuation, Features, SVR, MRA, Valuation Model for Smart Cities and Urban

Abstract. Determination of real estate value plays a very critical role in economic development and basic needs of people. Increasing demand for real estate together with population growth is making it difficult to determine real estate value. In applications where real estate is the main subject, such as urban activities, smart cities and urbanization, urban information system and valuation systems, model-based value estimations are essential for effective land/real estate policy. The type of real estate and impact degree of features depending on the type should be known as well as value estimation. It will be beneficial to follow a method that both determines the real estate value and factor impact degree. With the studies to be carried out using such methods, both region-specific valuation models can be created and the model is established with the optimum variable. This paper aimed to determine real estate value by using Support Vector Regression (SVR) and Multi Regression Analysis (MRA) methods for effective real estate management. Besides, both methods were examined by revealing the impact degrees of features that affect the value. The methods were applied to 319 parcels in Konya. For each parcel, 31 land features and market values were collected. The parcel data collected since 2018 were included in the models. From the results, the RBF-SVR model reached the highest R2 value with 0.88, while the MRA model reached 0.86.