Analyzing the Impact of Visual and Listing Features in Real Estate Listings

Bayrak S. N., ALPTEKİN G., ORMAN G. K., Arslan A.

Workshops of the International Conference on Computational Science and Its Applications, ICCSA 2025, İstanbul, Turkey, 30 June - 03 July 2025, vol.15886 LNCS, pp.274-285, (Full Text) identifier

  • Publication Type: Conference Paper / Full Text
  • Volume: 15886 LNCS
  • Doi Number: 10.1007/978-3-031-97576-9_18
  • City: İstanbul
  • Country: Turkey
  • Page Numbers: pp.274-285
  • Keywords: CTR, feature selection, Image feature extraction, image quality, machine learning, real estate
  • Galatasaray University Affiliated: Yes

Abstract

This study investigates the relationship between the visual and listing features of real estate listings and their corresponding Click-Through Rates (CTR). Several machine learning methods were used to identify the key attributes influencing CTR, including Decision Tree, Linear Regression, Random Forrest and XG-Boost. The models were trained and evaluated using a dataset comprising 3000 real estate listings. Visual features were extracted from listings’ cover images, while listing-related features – such as price, number of rooms, and district, were derived. Our findings indicate that visual image features and listing features have impact on CTR. In the study we also isolate the visual image and listing features and observed their individual contribution to CTR.