Details
| Translated title of the contribution | Aggregation and weighting technique of different real estate valuation data in regions with few transactions by means of variance component estimation |
|---|---|
| Original language | German |
| Pages (from-to) | 123-136 |
| Number of pages | 14 |
| Journal | AVN Allgemeine Vermessungs-Nachrichten |
| Volume | 124 |
| Issue number | 5 |
| Publication status | Published - 2017 |
Abstract
In Germany, real estate valuation is realized by three standardized methods, among which the approach based on sales comparison is considered as being closest to the real estate market. With this approach, regression analysis is frequently used to solve the linear relationship. The regression model, however, normally requires approximately 15 purchases per independent variable for an accurate estimate in real estate valuation. This causes a problem for areas with few transactions in which only 10 to 30 purchases are available. To solve this problem, the current paper presents a mathematical-statistical model for an accurate estimation of regression coefficients in areas with few transactions. This aim is realized by using additional market data, which are combined with the available purchase cases. Challenges with this approach are the acquisition of the additional market data and the optimal combination of the data in the adjustment model. In this paper, survey data from real estate experts and offer prices are used as additional market data. The combination of these data with the purchase cases is carried out by means of variance component estimation (VCE). The quality assessment of the various market data shows a good conformity between the purchase cases and the experts' survey data. The offer prices differ from the other two datasets. The investigation concerning the VCE for the weighting of the different datasets within the adjustment model has two results: VCE yields more precise estimates of the regression coefficients, but this does not allow for a better prediction of the market values.
ASJC Scopus subject areas
- Social Sciences(all)
- Geography, Planning and Development
- Engineering(all)
- Civil and Structural Engineering
- Earth and Planetary Sciences(all)
- Earth and Planetary Sciences (miscellaneous)
Cite this
- Standard
- Harvard
- Apa
- Vancouver
- BibTeX
- RIS
In: AVN Allgemeine Vermessungs-Nachrichten, Vol. 124, No. 5, 2017, p. 123-136.
Research output: Contribution to journal › Article › Research › peer review
}
TY - JOUR
T1 - Aggregation und gewichtung von unterschiedlichen wertermittlungsdaten in kaufpreisarmen lagen mittels varianzkomponentenschatzung
AU - Dorndorf, Alexander
AU - Soot, Matthias
AU - Weitkamp, Alexandra
AU - Alkhatib, Hamza
N1 - Copyright: Copyright 2017 Elsevier B.V., All rights reserved.
PY - 2017
Y1 - 2017
N2 - In Germany, real estate valuation is realized by three standardized methods, among which the approach based on sales comparison is considered as being closest to the real estate market. With this approach, regression analysis is frequently used to solve the linear relationship. The regression model, however, normally requires approximately 15 purchases per independent variable for an accurate estimate in real estate valuation. This causes a problem for areas with few transactions in which only 10 to 30 purchases are available. To solve this problem, the current paper presents a mathematical-statistical model for an accurate estimation of regression coefficients in areas with few transactions. This aim is realized by using additional market data, which are combined with the available purchase cases. Challenges with this approach are the acquisition of the additional market data and the optimal combination of the data in the adjustment model. In this paper, survey data from real estate experts and offer prices are used as additional market data. The combination of these data with the purchase cases is carried out by means of variance component estimation (VCE). The quality assessment of the various market data shows a good conformity between the purchase cases and the experts' survey data. The offer prices differ from the other two datasets. The investigation concerning the VCE for the weighting of the different datasets within the adjustment model has two results: VCE yields more precise estimates of the regression coefficients, but this does not allow for a better prediction of the market values.
AB - In Germany, real estate valuation is realized by three standardized methods, among which the approach based on sales comparison is considered as being closest to the real estate market. With this approach, regression analysis is frequently used to solve the linear relationship. The regression model, however, normally requires approximately 15 purchases per independent variable for an accurate estimate in real estate valuation. This causes a problem for areas with few transactions in which only 10 to 30 purchases are available. To solve this problem, the current paper presents a mathematical-statistical model for an accurate estimation of regression coefficients in areas with few transactions. This aim is realized by using additional market data, which are combined with the available purchase cases. Challenges with this approach are the acquisition of the additional market data and the optimal combination of the data in the adjustment model. In this paper, survey data from real estate experts and offer prices are used as additional market data. The combination of these data with the purchase cases is carried out by means of variance component estimation (VCE). The quality assessment of the various market data shows a good conformity between the purchase cases and the experts' survey data. The offer prices differ from the other two datasets. The investigation concerning the VCE for the weighting of the different datasets within the adjustment model has two results: VCE yields more precise estimates of the regression coefficients, but this does not allow for a better prediction of the market values.
KW - Expert knowledge
KW - Regions with few transactions
KW - Regression analysis
KW - Variance component estimation
UR - http://www.scopus.com/inward/record.url?scp=85019863887&partnerID=8YFLogxK
M3 - Artikel
AN - SCOPUS:85019863887
VL - 124
SP - 123
EP - 136
JO - AVN Allgemeine Vermessungs-Nachrichten
JF - AVN Allgemeine Vermessungs-Nachrichten
SN - 0002-5968
IS - 5
ER -