近年國際主要都會區普遍面臨住宅價格高漲、租金負擔上升與價租關係分化等問題，使住宅市場同時涉及資產配置與居住可負擔性。從不動產市場理論來看，住宅買賣市場較接近資產市場，反映所有權價值、區位稀缺性、資本化效果與未來增值預期；租賃市場則較接近空間市場，主要反映當期居住服務、生活便利性與承租者支付能力。因此，房價與租金即使受到相似住宅屬性與區位條件影響，其定價邏輯仍可能存在差異。本研究以2020年至2025年臺北市實價登錄資料為基礎，分別建構住宅買賣市場與租賃市場資料集，並結合捷運站、學校與公園等公共設施資料，建立住宅結構、內部使用、外部管理，以及區位與鄰里環境等變數。方法上，本研究採用CatBoost 梯度提升演算法建立房價與租金預測模型，並結合SHAP（Shapley Additive exPlanations）可解釋性方法，比較兩個市場的主要影響因素、非線性關係與市場轉折點。
實證結果顯示，房價模型主要受區位與資產屬性影響，其中行政區與至最近捷運站距離為重要變數，顯示買賣市場較能反映區位資本化與資產價值評估。相較之下，租金模型較受建物面積、屋齡、住宅類型與室內設備等居住使用條件影響，反映租賃市場更貼近當期居住服務需求。進一步分析非線性效果可發現，至最近捷運站距離對房價與租金的轉折點分別約為593公尺與165公尺；屋齡轉折點則分別約為35年與44年，顯示兩個市場對交通可及性與建物老化的評價方式並不相同。整體而言，本研究發現臺北市住宅買賣市場較偏向資產市場邏輯，房價主要反映區位稀缺性、資本化效果與未來價值預期；租賃市場則較接近空間市場邏輯，租金主要反映當期居住效用與承租者支付能力限制。本文透過可解釋機器學習方法，補充傳統特徵價格模型對非線性關係與市場異質性掌握不足之處，並提供理解臺北市房價與租金定價差異的實證依據。

In recent years, major metropolitan areas worldwide have faced rising housing prices, increasing rental burdens, and growing divergence between housing prices and rents. These issues indicate that housing markets involve not only residential needs but also asset allocation and housing affordability. From the perspective of real estate market theory, the housing sales market is more closely related to the asset market, reflecting ownership value, locational scarcity, capitalization effects, and expectations of future appreciation. In contrast, the rental market is closer to the space market, mainly reflecting current residential services, living convenience, and tenants’ affordability. Therefore, although similar housing attributes and locational conditions may influence housing prices and rents, their pricing mechanisms may differ. Based on Taipei City's real estate transaction registration data from 2020 to 2025, this study constructs separate datasets for the housing sales and rental markets. Public facility data, including MRT stations, schools, and parks, are further incorporated to construct variables related to housing structure, internal housing features, external management characteristics, and locational and neighborhood environments. Methodologically, this study employs the CatBoost gradient boosting algorithm to build predictive models for housing prices and rents, and applies SHAP (Shapley Additive exPlanations) to compare the key determinants, nonlinear relationships, and market turning points of the two markets.
The empirical results show that the housing price model is mainly influenced by locational and asset-related attributes, with administrative district and distance to the nearest MRT station serving as important variables. This indicates that the sales market reflects locational capitalization and asset value assessment. In contrast, the rent model is more strongly affected by residential use conditions, such as building area, building age, housing type, and indoor equipment, suggesting that the rental market is more closely related to current residential service demand. Further analysis of nonlinear effects shows that the turning points for distance to the nearest MRT station are approximately 593 meters for housing prices and 165 meters for rents. The turning points for building age are approximately 35 years and 44 years, respectively, indicating that the two markets evaluate transportation accessibility and building depreciation differently. Overall, this study finds that Taipei City's housing sales market is more consistent with asset-market logic, in which housing prices mainly reflect locational scarcity, capitalization effects, and expectations of future value. By contrast, the rental market is closer to space-market logic, in which rents mainly reflect current residential utility and tenants' affordability constraints. By applying explainable machine learning, this study complements traditional hedonic price models by better capturing nonlinear relationships and market heterogeneity. It provides empirical evidence for understanding the differences between housing price and rent formation in Taipei City.

Acuña, R. (2023). End of the line: The impact of new suburban rail stations on housing prices. The Journal of Transport and Land Use 16(1), 67–86.
Agostini, C. A. and G. A. Palmucci (2008). The anticipated capitalisation effect of a new metro line on housing prices. Fiscal Studies 29(2), 233–256.
Bae, C.-H. C., M.-J. Jun, and H.-W. Park (2003). The impact of seoul’s subway line 5 on residential property values. Transport Policy 10(2), 85–94.
Benjamin, J. D. and G.S.Sirmans (1996). Mass transportation, apartment rent and property values. Journal of Real Estate Research 12(1), 1–8.
Biecek, P. and T. Burzykowski (2021). Explanatory Model Analysis: Explore, Explain, and Examine Predictive Models. Boca Raton: CRC Press.
Breiman, L. (2001). Random forests. Machine Learning 45(1), 5–32.
Chen, T. and C. Guestrin (2016). XGBoost: A scalable tree boosting system. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. Association for Computing Machinery.
Chin, T. L. and K. W. Chau (2003). A critical review of literature on the hedonic price model. International Journal for Housing Science and Its Applications 27(2), 145–165.
Des Rosiers, F. and M. Thériault (1996). Rental amenities and the stability of hedonic prices: A comparative analysis of five market segments. Journal of Real Estate Research 12(1), 17–36.
DiPasquale, D. and W. C. Wheaton (1992). The markets for real estate assets and space: A conceptual framework. Journal of the American Real Estate and Urban Economics Association 20(1), 181–197.
Doshi-Velez, F. and B. Kim (2017). Towards a rigorous science of interpretable machine learning. arXiv preprint arXiv:1702.08608.
Friedman, J. H. (2001). Greedy function approximation: A gradient boosting machine. The Annals of Statistics 29(5), 1189–1232.
Goodman, A. C. and T. G. Thibodeau (1995). Age-related heteroskedasticity in hedonic house price equations. Journal of Housing Research 6(1), 25–42.
Halket, J., L. Nesheim, and F. Oswald (2015). The housing stock, housing prices, and user costs: The roles of location, structure and unobserved quality. cemmap working paper CWP73/15, Institute for Fiscal Studies.
Hartono, F.W., Muljono, and A.Z.Fanani(2024). Improving the accuracy of house price prediction using catboost regression with random search hyperparameter tuning: A comparative analysis. Advance Sustainable Science, Engineering and Technology 6(3).
Himmelberg, C., C. Mayer, and T. Sinai (2005). Assessing high house prices: Bubbles, fundamentals and misperceptions. The Journal of Economic Perspectives 19(4), 67–92.
Kahveci, M. and E. Sabaj (2017). Determinant of housing rents in urban albania: An empirical hedonic price application with nsa survey data. Eurasian Journal of Economics and Finance 5(2), 51–65.
Lee, B. S., E. C. Chung, and Y. H. Kim (2005). Dwelling age, redevelopment, and housing prices: The case of apartment complexes in seoul. Journal of Real Estate Finance and Economics 30(1), 55–80.
Li, Y., R. An, Y.Liu, Y.Liu, S.Jiang, Z.Lu, and L.Zhang(2025). Exploring nonlinear relationships between street quality and housing prices: A comparison of sales and rentals. Journal of Urban Management.
Loewenstein, L. and J. Meyer (2024). Comparing two house-price booms. Economic Commentary 2024-04, Federal Reserve Bank of Cleveland. Available at: https://www.clevelandfed.org/publications/economic-commentary/2024/ec-202404-comparing-two-house-price-booms.
Lundberg, S. M., G. Erion, H. Chen, A. DeGrave, J. M. Prutkin, B. Nair, R. Katz, J. Himmelfarb, N.Bansal, and S.-I.Lee(2020). From local explanations to global understanding with explainable ai for trees. Nature Machine Intelligence 2(1), 56–67.
Lundberg, S. M. and S.-I. Lee (2017). A unified approach to interpreting model predictions.
Mayo, S. K.(1981). Theory and estimation in the economics of housing demand. Journal of Urban Economics 10, 95–116.
Molnar, C. (2022). Interpretable Machine Learning: A Guide for Making Black Box Models Explainable. Independently published.
Prokhorenkova, L., G. Gusev, A. Vorobev, A. V. Dorogush, and A. Gulin (2018). CatBoost: unbiased boosting with categorical features. arXiv preprint arXiv:1706.09516.
Rennert, L. (2022). A meta-analysis of the impact of rail stations on property values: Applying a transit planning lens. Transportation Research Part A: Policy and Practice 163, 165–180.
Ribeiro, M. T., S. Singh, and C. Guestrin (2016). “Why Should I Trust You?”: Explaining the predictions of any classifier. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp. 1135–1144. ACM.
Sinai, T. and N. S. Souleles (2005). Owner-occupied housing as a hedge against rent risk. The Quarterly Journal of Economics 120(2), 763–789.
Sirmans, G. S., D. A. Macpherson, and E. N. Zietz (2005). The composition of hedonic pricing models. Journal of Real Estate Literature 13(1), 1–44.
Tseng, C.-C., C.-O. Chang, and C.-I. Hua (2006). The dynamic relationship of rents and prices among Taipei housing spatial submarkets. Journal of Housing Studies 14(2), 27–49. In Chinese.
Wang, W.-K. and I.-C. Tsai (2022). House age and housing prices: A viewpoint of the optimal time for land redevelopment. International Journal of Strategic Property Management 26(3), 172–187.
Won, J. and J.-S. Lee (2018). Investigating how the rents of small urban houses are determined: Using spatial hedonic modeling for urban residential housing in seoul. Sustainability 10(1), 31.
Yang, B.-Y. (2018). Why is taipei housing expected appreciation rising fast? — Using price rent ratio to discuss the differences between consumption value and investment value. Master’s thesis, National Chengchi University, Taipei, Taiwan.
Zhou, F., H.-M. Zhou, Z.-H. Yang, and L.-H. Yang (2019). A 2-stage strategy for non-stationary signal prediction and recovery using iterative filtering and neural network. Journal of Computer Science and Technology 34(2), 318–338.