Locally Regularized Linear Regression in the Valuation of Real Estate

Mariusz Kubus ()

Statistics in Transition new series, 2016, vol. 17, issue 3, 515-524

Abstract: Regression methods are used for the valuation of real estate in the comparative approach. The basis for the valuation is a data set of similar properties, for which sales transactions were concluded within a short period of time. Large and standardized databases, which meet the requirements of the Polish Financial Supervision Authority, are created in Poland and used by the banks involved in mortgage lending, for example. We assume that in the case of large data sets of transactions, it is more advantageous to build local regression models than a global model. Additionally, we propose a local feature selection via regularization. The empirical research carried out on three data sets from real estate market confirmed the effectiveness of this approach. We paid special attention to the model quality assessment using cross-validation for estimation of the residual standard error.

Keywords: large transactional data; local regression; feature selection; regularization; cross-validation (search for similar items in EconPapers)
Date: 2016
References: View references in EconPapers View complete reference list from CitEc
Citations: View citations in EconPapers (1)

Downloads: (external link)
http://index.stat.gov.pl/repec/files/csb/stintr/csb_stintr_v17_2016_i3_n9.pdf (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:csb:stintr:v:17:y:2016:i:3:p:515-524

Access Statistics for this article

Statistics in Transition new series is currently edited by Włodzimierz Okrasa

More articles in Statistics in Transition new series from Główny Urząd Statystyczny (Polska) Contact information at EDIRC.
Bibliographic data for series maintained by Beata Witek ( this e-mail address is bad, please contact ).