guest ::
| Home > Conference materials > Papers > A Robustified Metalearning Procedure for Regression Estimators |
Original title:
A Robustified Metalearning Procedure for Regression Estimators
Authors: Kalina, Jan ; Neoral, A.
Document type: Papers
Conference/Event: International Days of Statistics and Economics /13./, Prague (CZ), 20190905
Year: 2019
Language: eng
Abstract: Metalearning represents a useful methodology for selecting and recommending a suitable algorithm or method for a new dataset exploiting a database of training datasets. While metalearning is potentially beneficial for the analysis of economic data, we must be aware of its instability and sensitivity to outlying measurements (outliers) as well as measurement errors. The aim of this paper is to robustify the metalearning process. First, we prepare some useful theoretical tools exploiting the idea of implicit weighting, inspired by the least weighted squares estimator. These include a robust coefficient of determination, a robust version of mean square error, and a simple rule for outlier detection in linear regression. We perform a metalearning study for recommending the best linear regression estimator for a new dataset (not included in the training database). The prediction of the optimal estimator is learned over a set of 20 real datasets with economic motivation, while the least squares are compared with several (highly) robust estimators. We investigate the effect of variable selection on the metalearning results. If the training as well as validation data are considered after a proper robust variable selection, the metalearning performance is improved remarkably, especially if a robust prediction error is used.
Keywords: computational statistics; model choice; robustness; variable selection
Project no.: GA17-07384S (CEP)
Funding provider: GA ČR
Host item entry: The 13th International Days of Statistics and Economics Conference Proceedings, ISBN 978-80-87990-18-6
Note: Související webová stránka: https://msed.vse.cz/msed_2019/sbornik/toc.html
Institution: Institute of Computer Science AS ČR (web)
Document availability information: Fulltext is available in the digital repository of the Academy of Sciences.
Original record: http://hdl.handle.net/11104/0300999
Permalink: http://www.nusl.cz/ntk/nusl-407883
The record appears in these collections:
Research > Institutes ASCR > Institute of Computer Science
Conference materials > Papers
Authors: Kalina, Jan ; Neoral, A.
Document type: Papers
Conference/Event: International Days of Statistics and Economics /13./, Prague (CZ), 20190905
Year: 2019
Language: eng
Abstract: Metalearning represents a useful methodology for selecting and recommending a suitable algorithm or method for a new dataset exploiting a database of training datasets. While metalearning is potentially beneficial for the analysis of economic data, we must be aware of its instability and sensitivity to outlying measurements (outliers) as well as measurement errors. The aim of this paper is to robustify the metalearning process. First, we prepare some useful theoretical tools exploiting the idea of implicit weighting, inspired by the least weighted squares estimator. These include a robust coefficient of determination, a robust version of mean square error, and a simple rule for outlier detection in linear regression. We perform a metalearning study for recommending the best linear regression estimator for a new dataset (not included in the training database). The prediction of the optimal estimator is learned over a set of 20 real datasets with economic motivation, while the least squares are compared with several (highly) robust estimators. We investigate the effect of variable selection on the metalearning results. If the training as well as validation data are considered after a proper robust variable selection, the metalearning performance is improved remarkably, especially if a robust prediction error is used.
Keywords: computational statistics; model choice; robustness; variable selection
Project no.: GA17-07384S (CEP)
Funding provider: GA ČR
Host item entry: The 13th International Days of Statistics and Economics Conference Proceedings, ISBN 978-80-87990-18-6
Note: Související webová stránka: https://msed.vse.cz/msed_2019/sbornik/toc.html
Institution: Institute of Computer Science AS ČR (web)
Document availability information: Fulltext is available in the digital repository of the Academy of Sciences.
Original record: http://hdl.handle.net/11104/0300999
Permalink: http://www.nusl.cz/ntk/nusl-407883
The record appears in these collections:
Research > Institutes ASCR > Institute of Computer Science
Conference materials > Papers
Record created 2019-11-25, last modified 2023-12-06