Mass Appraisal Based on Geographic Information Systems and Explainable Artificial Intelligence Using TUCBS Data

Öz

This study presents a mass appraisal approach that integrates TUCBS data with Geographic Information Systems (GIS) and artificial intelligence methods to estimate residential property values in the Yenimahalle and Keçiören districts of Ankara. A total of 37,095 property sales records obtained from Endeksa were enriched with spatial layers—including Digital Elevation Model (DEM), transportation networks, education and health facilities, and Point of Interest (POI) data—acquired from TUCBS, resulting in a comprehensive dataset consisting of 34,272 residential sales records and 56 variables. Following exploratory data analysis and feature engineering, variable importance was determined using the Permutation Feature Importance (PFI) method, and variables with low contribution were removed. Eight machine learning models (Random Forest, Extra Trees, Bagging, Gradient Boosting, AdaBoost, XGBoost, LightGBM, CatBoost) were optimized using GridSearchCV and evaluated based on R², RMSE, and MAE metrics. The results show that XGBoost, LightGBM, and Random Forest achieved the highest accuracy (R² ≈ 0.91), while LightGBM provided the best balance between accuracy and computation time. The decision-making process of the models was clarified through Explainable Artificial Intelligence (XAI) techniques, with SHAP analysis revealing the directional and quantitative contributions of each feature to the predictions. Findings indicate that total floor count, gross area, building age, and floor level are the most influential determinants of residential property value. Estimated values were mapped spatially using 500 m² hexagonal grids and published via ArcGIS Server. The study demonstrates that the integration of ML–GIS–XAI can provide a transparent, scalable, and data-driven mass appraisal infrastructure, offering an effective decision-support tool for banking, property taxation, expropriation, and urban planning.

Anahtar Kelimeler

• Mass Appraisal
• Geographic Information Systems
• Machine Learning
• Explainable Artificial Intelligence
• TUCBS

TUCBS Verileriyle Coğrafi Bilgi Sistemleri ve Açıklanabilir Yapay Zekâ Tabanlı Toplu Değerleme

Öz

Bu çalışma, TUCBS verilerinin Coğrafi Bilgi Sistemleri (CBS) ve yapay zekâ yöntemleri kullanılarak Ankara’nın Yenimahalle ve Keçiören ilçelerinde konut değerlerini tahmin etmeye yönelik bir toplu değerleme yaklaşımı sunmaktadır. Endeksa’dan elde edilen 37.095 konut satış verisi; TUCBS’den temin edilen sayısal yükseklik modeli (DEM), ulaşım, eğitim, sağlık tesisleri ve ilgi noktaları (POI) gibi katmanlar analiz edilerek mekânsal verilerle zenginleştirilmiş ve 34.272 konut satış kaydı ile 56 değişkenden oluşan kapsamlı bir veri seti oluşturulmuştur. Keşifsel veri analizi ve özellik mühendisliği sonrasında Permütasyon Özellik Önemi (PFI) yöntemiyle değişken önemi belirlenmiş, katkısı düşük değişkenler çıkarılmıştır. Sekiz makine öğrenmesi modeli (Random Forest, Extra Trees, Bagging, Gradient Boosting, AdaBoost, XGBoost, LightGBM, CatBoost) GridSearchCV kullanılarak optimize edilmiş ve performansları R², RMSE ve MAE metrikleri üzerinden değerlendirilmiştir. Sonuçlar, XGBoost, LightGBM ve Random Forest modellerinin en yüksek doğruluğu sağladığını (R² ≈ 0.91), LightGBM’in ise doğruluk-zaman dengesi bakımından en uygun model olduğunu göstermiştir. Modelin karar mekanizması; açıklanabilir yapay zekâ (XAI) teknikleriyle şeffaflaştırılmış, SHAP analizi ile her bir özelliğin tahmine yönlü ve nicel etkilerini açıklamıştır. Bulgular; toplam kat sayısı, brüt alan, bina yaşı ve bulunduğu kat değişkenlerinin konut değeri üzerinde en belirleyici faktörler olduğunu göstermektedir. Elde edilen değerler 500 m²lik altıgen ızgara yöntemiyle mekânsal olarak haritalandırılıp ArcGIS Server üzerinden servis edilmiştir. Çalışma, ML-GIS-XAI entegrasyonuyla şeffaf, ölçeklenebilir ve veri odaklı bir toplu değerleme altyapısı oluşturulabileceğini ve bu yaklaşımın bankacılık, emlak vergisi, kamulaştırma ve kentsel planlama gibi alanlarda etkili bir karar destek aracı sağlayabileceğini göstermektedir.

Anahtar Kelimeler

• Toplu değerleme
• Coğrafi Bilgi Sistemleri
• Makine Öğrenmesi
• Açıklanabilir Yapay Zekâ
• TUCBS

Kaynakça

1. Adadi, A., & Berrada, M. (2018). Peeking Inside the Black-Box: A Survey on Explainable Artificial Intelligence (XAI). IEEE Access, 6, 52138-52160. https://doi.org/10.1109/ ACCESS.2018.2870052
2. Botchkarev, A. (2019). A New Typology Design of Performance Metrics to Measure Errors in Machine Learning Regression Algorithms. Interdisciplinary Journal of Information, Knowledge, and Management, 14, 045-076. https://doi.org/10.28945/4184
3. Bühlmann, P. (2012). Bagging, Boosting and Ensemble Methods. Içinde J. E. Gentle, W. K. Härdle, & Y. Mori (Ed.), Handbook of Computational Statistics (ss. 985-1022). Springer Berlin Heidelberg. https://doi.org/10.1007/978-3-642-21551-3_33
4. Čeh, M., Kilibarda, M., Lisec, A., & Bajat, B. (2018). Estimating the Performance of Random Forest versus Multiple Regression for Predicting Prices of the Apartments. ISPRS International Journal of Geo-Information, 7(5), 168. https://doi.org/10.3390/ ijgi7050168
5. Chawla, A. (2023). A Visual and Overly Simplified Guide To Bagging and Boosting. https://blog.dailydoseofds.com/p/a-visual-and-overly-simplified-guide-4b1 Chen, T., & Guestrin, C. (2016). XGBoost: A Scalable Tree Boosting System.
6. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 785-794. https://doi.org/10.1145/2939672.2939785 Coğrafi İstatistik Portalı. (2025). Coğrafi İstatistik Portalı. https://cip.tuik.gov.tr/
7. Çılgın, C., Gökçen, H., & Gazi University, Ankara (Turkiye). (2023). Machine learning methods for prediction real estate sales prices in Turkey. Revista de La Construcción, 22(1), 163-177. https://doi.org/10.7764/RDLC.22.1.163
8. Dimopoulos, T., & Bakas, N. (2019). Sensitivity Analysis of Machine Learning Models for the Mass Appraisal of Real Estate. Case Study of Residential Units in Nicosia, Cyprus. Remote Sensing, 11(24), 3047. https://doi.org/10.3390/rs11243047

9. Freund, Y., & Schapire, R. E. (1996). Experiments with a New Boosting Algorithm. Gabrielli, L., & French, N. (2021). Pricing to market: Property valuation methods – a practical review. Journal of Property Investment & Finance, 39(5), 464-480. https:// doi.org/10.1108/JPIF-09-2020-0101
10. Gao, Q., Shi, V., Pettit, C., & Han, H. (2022). Property Valuation Using Machine Learning Algorithms On Statistical Areas In Greater Sydney, Australia. Land Use Policy, 123, 106409. https://doi.org/10.1016/j.landusepol.2022.106409
11. Guliker, E., Folmer, E., & Van Sinderen, M. (2022). Spatial Determinants of Real Estate Appraisals in The Netherlands: A Machine Learning Approach. ISPRS International Journal of Geo-Information, 11(2), 125. https://doi.org/10.3390/ijgi11020125
12. Hjort, A., Scheel, I., Sommervoll, D. E., & Pensar, J. (2024). Locally interpretable tree boosting: An application to house price prediction. Decision Support Systems, 178, 114106. https://doi.org/10.1016/j.dss.2023.114106
13. IAAO. (2018). Standard on Mass Appraisal of Real Property [Standart]. IAAO. https:// www.iaao.org/industry-data/iaao-technical-standards/
14. İban, M. C. (2021). Taşınmaz Mal Değeri Kestiriminde Topluluk Algoritmalarının Doğruluk Analizi. 47.
15. Ja’afar, N. S., Mohamad, J., & Ismail, S. (2021). Machine Learning For Property Price Prediction And Price Valuation: A Systematic Literature Review. Planning Malasia, 19. https://doi.org/10.21837/pm.v19i17.1018
16. Krämer, B., Stang, M., Doskoč, V., Schäfers, W., & Friedrich, T. (2023). Automated valuation models: Improving model performance by choosing the optimal spatial training level. Journal of Property Research, 40(4), 365-390. https://doi.org/10.1080 /09599916.2023.2206823
17. Lin, W., Shi, Z., Wang, Y., & Yan, T. H. (2023). Unfolding Beijing in a Hedonic Way. Computational Economics, 61(1), 317-340. https://doi.org/10.1007/s10614-021-10209-3 Lundberg, S. M., Erion, G., & Chen, H. (2019). Explainable AI for Trees: From Local Explanations to Global Understanding (No. arXiv:1905.04610). arXiv. https://doi. org/10.48550/arXiv.1905.04610
18. Michae Dellstad. (2018). Comparing three machine learning algorithms in the task of appraising commercial real estate (Tez). KTH Royal Institute Of Technology School Of Electrical Engineering And Computer Science.
19. Numan, J. A. A., & Yusoff, I. M. (2024). Identifying the Current Status of Real Estate Appraisal Methods. Real Estate Management and Valuation, 32(4), 12-27. https:// doi.org/10.2478/remav-2024-0032
20. Peng, Z., Huang, Q., & Han, Y. (2019). Model Research on Forecast of Second-Hand House Price in Chengdu Based on XGboost Algorithm. 2019 IEEE 11th International Conference on Advanced Infocomm Technology (ICAIT), 168-172. https://doi. org/10.1109/icait.2019.8935894
21. Perçem, E. (2025). Proptech Odaklı Toplu Değerleme: TUCBS Verilerine Dayalı Yapay Zekâ ve Coğrafi Bilgi Sistemleri Perspektifi (Coğrafi Bilgi Sistemi Uzmanlık Tezi).
22. Plevris, V., Solorzano, G., Bakas, N., & Ben Seghier, M. (2022). Investigation of performance metrics in regression analysis and machine learning-based prediction models. 8th European Congress on Computational Methods in Applied Sciences and Engineering. 8th European Congress on Computational Methods in Applied Sciences and Engineering. https://doi.org/10.23967/eccomas.2022.155
23. Rampini, L., & Re Cecconi, F. (2022). Artificial intelligence algorithms to predict Italian real estate market prices. Journal of Property Investment & Finance, 40(6), 588-611. https://doi.org/10.1108/JPIF-08-2021-0073
24. Renigier-Biłozor, M., Źróbek, S., Walacik, M., Borst, R., Grover, R., & d’Amato, M. (2022). International acceptance of automated modern tools use must-have for sustainable real estate market development. Land Use Policy, 113, 105876. https:// doi.org/10.1016/j.landusepol.2021.105876
25. Rolli, C. S. (2020). Zillow Home value Prediction Using XGBOOST. 47.
26. Stang, M. (2023). Real Estate Valuation in the Age of Artificial Intelligence – Modern Machine Learning Algorithms and their Application in Property Appraisal [Tez]. Submitted to the Faculty of Business, Economics, and Management Information Systems at the University of Regensburg.
27. Steurer, M., Hill, R. J., & Pfeifer, N. (2021). Metrics for evaluating the performance of machine learning based automated valuation models. Journal of Property Research, 38(2), 99-129. https://doi.org/10.1080/09599916.2020.1858937
28. TDUB. (2025). Değerleme Sektörü 2025 Yılı 1. Çeyrek Raporu (Analiz Raporu). MKK Gayrimenkul Bilgi Merkezi A.Ş. https://tdub.org.tr/sayfa/degerleme-sektoru-ozet-verileri Torres-Pruñonosa, J., García-Estévez, P., & Prado-Román, C. (2021). Artificial Neural Network, Quantile and Semi-Log Regression Modelling of Mass Appraisal in Housing. Mathematics, 9(7), 783. https://doi.org/10.3390/math9070783
29. Vargason, D. (2019). Data Mining Techniques for Predicting Real Estate Trends.
30. Wang, D., & Li, V. J. (2019). Mass Appraisal Models of Real Estate in the 21st Century: A Systematic Literature Review. Sustainability, 11(24), 7006. https://doi.org/10.3390/ su11247006
31. Xu, X., & Zhang, Y. (2021). House price forecasting with neural networks. Intelligent Systems with Applications, 12, 200052. https://doi.org/10.1016/j.iswa.2021.200052
32. Yagmur, A. (2025). Real Estate Valuation Decision-Making System Using Machine Learning and Geospatial Data.
33. Yang, H., Qiu, R., & Chen, W. (Ed.). (2020). Smart Service Systems, Operations Management, and Analytics: Proceedings of the 2019 INFORMS International Conference on Service Science. Springer International Publishing. https://doi. org/10.1007/978-3-030-30967-1
34. Yoshida, T., Murakami, D., & Seya, H. (2022). Spatial Prediction of Apartment Rent using Regression-Based and Machine Learning-Based Approaches with a Large Dataset. The Journal of Real Estate Finance and Economics, 69(1), 1-28. https://doi. org/10.1007/s11146-022-09929-6
35. Zhang, H., Li, Y., & Branco, P. (2024). Describe the house and I will tell you the price: House price prediction with textual description data. Natural Language Engineering, 30(4), 661-695. https://doi.org/10.1017/s1351324923000360