1{2 "record_id": "bc5c2c3c86f1f45909dcce75c8f09acff914dd28",3 "eligible": true,4 "short_ref": "An, Song, Jang & Ahn (2025)",5 "country": "South Korea",6 "city": "Busan",7 "property_type": "condominiums (apartments)",8 "price_type": "transaction price (log)",9 "n_obs": 61152,10 "sample_period": "2015",11 "data_source": "Song et al. (2021) hedonic dataset of condominium transactions",12 "methods_compared": [13 "OLS hedonic",14 "spatial lag regression (SLR)",15 "random forest",16 "deep neural network"17 ],18 "best_method": "random forest",19 "metric": "R2",20 "best_value": 0.941,21 "hedonic_value": null,22 "test_set": "holdout",23 "features_ai": "none",24 "incremental_ai_gain": null,25 "incremental_ai_metric": "",26 "marginal_effects": "SHAP",27 "notes": "70/30 train-test split. RF baseline RMSE 0.080 (log price) and R2 0.941 printed in appendix Table 8 (tuned RF 0.079/0.942). OLS and SLR accuracy appear only graphically in Fig. 3 (OLS R2 roughly 0.67, RMSE roughly 0.19 by eye), so hedonic_value is left null. Compares SHAP ranks with Gini importance and with OLS/SLR coefficient signs (Tables 5-6).",28 "confidence": "medium"29}