Vol-11,Issue-6,November - December 2026
Author(s): Xinyi Sun
Abstract: Housing valuation model selection and manual review require joint consideration of predictive accuracy, spatial applicability, and capacity constraints. Using 20,640 California census block-group observations and 1,824 Guangzhou housing records, this study conducts spatial stress tests and supplementary non-spatial review analyses. The results show that model rankings vary with spatial holdout design. Under a fixed feature specification, interquartile-range clipping reduces overall error but increases errors or underprediction bias for some sample groups. In the Guangzhou sample, under a nominal 10% review budget, the high-absolute-error capture rates of model-disagreement, training-only error-learning, and predicted-price-prioritization rules are 23.83%, 38.14%, and 41.31%, respectively, compared with 9.86% for random review. Rankings change under relative-error evaluation, partly because negative price predictions are identified. The findings highlight the importance of aligning model selection and review prioritization with deployment conditions, subgroup diagnostics, and loss objectives. The evidence is limited to offline prediction and screening; practical application requires further validation using target-market spatial and temporal information and actual review records.
Keywords: housing valuation; model risk governance; regional risk; resource allocation; spatial cross-validation.
Article Info: Received: 11 Oct 2026; Received in revised form: 03 Nov 2026; Accepted: 07 Nov 2026; Available online: 12 Nov 2026
| Total Views: 57 | Downloads: 3 | Page No: 85-110 | Download Full Text PDF |
Cite this Article:
For Authors
- Current Conferences
- RSS Feed Complete Issue

- Call for Papers

- Author Guidelines
- Policies
- Peer Review Process

- Privacy Policy

- Publication Policies and Ethics

- Open Access Policy
- Retraction Policy
- Indexing and Archiving
- Submit Manuscript Online
- How to Publish with Us?
- Current Issue

- Archive Issue
- Complete Issue
- Track Paper

- Download Formats
- Join Us
- Blog
Archiving
- Google Scholar
- Academia
- WorldCat
- PBN-Polish Scientific Bibliography
- The university Library-Aalborg University (Denmark)
- ResearchBib
- RootIndexing
- Jurn-Academic Articles, Chapters and Theses
- Scinapse
About IJELS
For Authors
License

This work is licensed under a Creative Commons Attribution 4.0 International License.
Social Plugin
Copyright @ 2026 IJELS(Infogain Publication)
