1\begin{tabular}{llllrrll}2\toprule3Study & country & best model & metric & best & hedonic benchmark & AI-derived features & marginal effects? \\4\midrule5Tarasov \& Dessoulavy-Sliwins & Poland & XGBoost & MAPE & 8.990 & -- & none & SHAP \\6Dewi, Ulama \& Susilaningrum & Indonesia & artificial neural netw & MAPE & 31.040 & 34.240 & none & coefficients \\7Amri \& Tularam (2012) & Australia & neural network & R2 & 0.454 & 0.369 & none & coefficients \\8Cui et al. (2024) & China & MGWR & R2 & 0.613 & 0.346 & street-view CNN & coefficients \\9Ezzeldin et al. (2025) & Saudi Arabia & random forest & R2 & 0.906 & 0.803 & none & coefficients \\10Solovev \& Pröllochs (2021) & Germany & deep neural network + & MAE & 0.021 & 0.157 & floor-plan & coefficients \\11Nor \& Hussein (2025) & Somalia & artificial neural netw & MAE & 1200.000 & -- & none & coefficients \\12Takizawa (2023) & Japan & OLS hedonic + GCN-deri & RMSE & 11280.000 & -- & floor-plan & coefficients \\13Gillespie, Lyons \& Hynes (20 & Ireland & OLS hedonic with Small & R2 & 0.890 & 0.890 & listing text (rule-based N & coefficients \\14Han, Woo \& Lee (2025) & South Korea & gradient boosting & R2 & 0.817 & 0.660 & street-view CNN & SHAP \\15Sa'at, Adi Maimun \& Idris (2 & Malaysia & ANN (one hidden layer, & RMSE & 0.005 & 0.049 & none & none \\16Geerts et al. (2025) & Multi: Usa & LightGBM (with coordin & MAPE & 0.138 & -- & none & SHAP \\17Embaye, Zereyesus \& Chen (20 & Uganda & boosting & other & 0.850 & 1.000 & none & coefficients \\18Štubňová, Urbaníková, Hudáko & Slovakia & ANN (Bayesian regulari & MAPE & 3.390 & 8.410 & none & coefficients \\19Deppner \& Cajias (2024) & Germany & XGBoost & R2 & 0.852 & 0.788 & none & coefficients \\20Changro (2026) & South Korea & XGBoost with text embe & R2 & -- & -- & listing text embeddings & SHAP \\21Tapia, Chavez-Garzon, Pezoa, & Chile & LightGBM (34 predictor & MAPE & 10.070 & 21.130 & listing-photo CNN (interio & SHAP \\22Almaas, Oust \& Pollestad (20 & Norway & XGBoost with human-lab & MAPE & 7.490 & 12.410 & interior-photo CNN & SHAP \\23Upadhye, Bhuravane \& Khanche & United States & XGBoost + CNN text emb & R2 & 0.920 & 0.710 & listing text embeddings & SHAP \\24Lorenz, Willwersch, Cajias \& & Germany & XGBoost & R2 & 0.925 & -- & none & PDP \\25Fu et al. (2019) & China & robust OLS hedonic wit & R2 & 0.724 & 0.724 & street-view CNN & coefficients \\26Tymchuk, Pylypenko \& Kicha ( & Ukraine & OLS with NN text-deriv & R2 & 0.889 & 0.884 & listing text embeddings & coefficients \\27Yang, Krenz, Qiu \& Li (2023) & United Kingdom & OLS hedonic, Model 0 ( & R2 & 0.496 & 0.496 & street-view CNN & coefficients \\28Wang (2026b) & United States & random forest & MAPE & 10.430 & -- & none & SHAP \\29Nor, Audu \& Mohamed (2025a) & Somalia & OLS hedonic regression & MAE & 29.380 & 29.380 & none & coefficients \\30Shen \& Ross (2020) & United States & OLS hedonic + text-der & R2 & 0.769 & 0.760 & listing text embeddings & coefficients \\31d'Amato (2004) & Italy & linear MRA (hedonic) & MAPE & 11.000 & 11.000 & none & coefficients \\32Hong, Choi \& Kim (2020) & South Korea & random forest & MAPE & 5.482 & 19.606 & none & none \\33Wang et al. (2026a) & China & GBDT & R2 & 0.470 & -- & street-view CNN & PDP \\34Zhong (2024) & United States & OLS multiple linear re & RMSE & 0.203 & 0.203 & none & coefficients \\35Lenaers, Boudt \& De Moor (20 & Belgium & XGBoost & MAPE & 10.330 & 11.590 & none & SHAP \\36Oliveira, Reyes-Bueno, Gonza & Brazil & XGBoost & MAPE & 8.960 & 13.900 & none & coefficients \\37Haque (2003) & United States & Jordan-Elman recurrent & R2 & 0.863 & -- & none & coefficients \\38An, Song, Jang \& Ahn (2025) & South Korea & random forest & R2 & 0.941 & -- & none & SHAP \\39Marimon \& Vergara (2026) & Philippines & linear hedonic regress & R2 & 0.547 & 0.547 & street-view CNN & coefficients \\40Liu, Chen, Orford, Tian \& Zo & United Kingdom & XGBoost & R2 & 0.760 & -- & none & SHAP \\41Fan, Teo \& Wan (2021) & Singapore & OLS hedonic with block & R2 & 0.919 & 0.919 & text sentiment & coefficients \\42Kagie \& van Wezel (2007) & Netherlands & boosted regression tre & MAPE & 9.380 & -- & none & PDP \\43Kim, Choi \& Lee (2025) & South Korea & random forest & MAPE & 5.928 & 12.227 & none & SHAP \\44Tajani, Di Liddo, Ranieri \& & Italy & evolutionary polynomia & R2 & 0.887 & -- & none & coefficients \\45\bottomrule46\end{tabular}47