Robust variable selection with application to quality of life research |
| |
Authors: | Andreas Alfons Wolfgang E. Baaske Peter Filzmoser Wolfgang Mader Roland Wieser |
| |
Affiliation: | 1.Department of Statistics and Probability Theory,Vienna University of Technology,Vienna,Austria;2.STUDIA-Schlierbach, Studienzentrum für internationale Analysen,Schlierbach,Austria;3.SPES Academy,Schlierbach,Austria |
| |
Abstract: | A large database containing socioeconomic data from 60 communities in Austria and Germany has been built, stemming from 18,000 citizens’ responses to a survey, together with data from official statistical institutes about these communities. This paper describes a procedure for extracting a small set of explanatory variables to explain response variables such as the cognition of quality of life. For better interpretability, the set of explanatory variables needs to be very small and the dependencies among the selected variables need to be low. Due to possible inhomogeneities within the data set, it is further required that the solution is robust to outliers and deviating points. In order to achieve these goals, a robust model selection method, combined with a strategy to reduce the number of selected predictor variables to a necessary minimum, is developed. In addition, this context-sensitive method is applied to obtain responsible factors describing quality of life in communities. |
| |
Keywords: | |
本文献已被 SpringerLink 等数据库收录! |
|