Robust Bayesian inference in STAR models with neighbourhood effects

详细信息    查看全文

• 作者：Asunció ; n Beamonte ; Pilar Gargallo ; Manuel Salvador
• 关键词：STAR ; Neighbourhood effects ; Bayesian inference ; Robustness ; MCMC ; Hedonic models ; Real estate market ; Importance sampling ; Rolling validation
• 刊名：Journal of Statistical Planning and Inference
• 出版年：2010
• 出版时间：November 2010
• 年：2010
• 卷：140
• 期：11
• 页码：3047-3057
• 全文大小：1513 K

文摘

In this paper we propose a robust Bayesian procedure of estimation, testing, validation and selection of spatio-temporal autoregressive models (STAR) with neighbourhood effects applied to the appraisal of dwelling prices. The methodology does not depend on asymptotic results and, unlike previously procedures proposed in the literature, takes into account the uncertainty associated to the estimation of the neighbourhood parameters of the model, giving more realism to the analysis. Moreover, a sequential algorithm to elaborate fast on-line forecast, is provided. The methodology is illustrated by means of a practical case of the real estate market of Zaragoza.