Application of GAM Additive Non-Linear Models to Estimate Real Estate Market Value

• Autorzy: Janusz Dąbrowski , Tomasz Adamczyk
• Języki publikacji: EN

Abstrakty

EN

Generalized additive models (Generalized Additive Models - GAM) have a relatively short history, and hence their use is difficult. Although Hastie and Tibshirani [1] and Schimek [3] provide a detailed interpretation of them, yet in professional practice these methods were not used. In the literature for valuers there was not found a single citation for the application of GAM models. The most characteristic for generalized additive models of data analysis is a graph of the expected values and sub-rests, thanks to it we can see the final match of cubic (third degree curve) concatenated to the final model. Residual statistics as in other methods allow us to assess the quality of the fit and predict stand-off values. Developers often use non-linear estimation by calculating the relationship between the price of a property and the time needed to sell it. A linear relationship is assumed a priori in multiple regression analysis of variance. Nonlinear estimation allows the optimal choice of function which binds variables and appoints parameters. This may be, among others, a function: logarithmic, exponential, or being the product of independent variables. Linear function is relatively simple to interpret, in contrast to the nonlinear function where predictors are smoothed by the function of the third degree, which allows for a better fit of the model to follow. Hastie and Tibshirani (1990) suggest that R-square value, calculated as a relative improvement of the overall variance for the final model, was considered as a measure of fit of the model.(fragment of text)

Słowa kluczowe

PL

Modele nieliniowe; Programowanie liniowe; Rynek nieruchomości

EN

Nonlinear models; Linear programming; Real estate market

• Czasopismo: Geomatics and Environmental Engineering
• Rocznik: 2010
• Numer: nr 4/2
• Strony: 55-62

Twórcy

autor

Janusz Dąbrowski

• The Bronisław Markiewicz State Higher School of Technology and Economics in Jarosław, Poland

autor

Tomasz Adamczyk

• AGH University of Science and Technology Kraków, Poland

Bibliografia

• Hastie T.J., Tibshirani R.J.: Generalized Additive Models. Chapman & Hall, New York 1990.
• Schimek M.G.: Smoothing and regression: Approaches, computations, and application. Wiley, New York 2000.
• McCullagh P., Nelder J.A. Generalized linear models. 2nd Ed., Chapman & Hall, New York 1989.