M-quantile regression shrinkage and selection via the lasso
M-quantile regression shrinkage and selection via the lasso
Standard regression analysis investigates the average behavior of a response variable, y, given a vector of predictors, x. However, in some cases, the mean does not give a complete picture of a distribution. Therefore, quantile regression analyzes how the q-th quantile of the conditional distribution of y given x varies withx, and M-quantile regression generalizes this idea through the use of influence functions. When dealing with a large number of predictors, selection of a subset of themimproves the interpretability of the model. Towards this end, in this paper, we introduce M-quantile regression with lasso regularization. This allows us to investigate the extreme behavior of y conditional on x and to shrink the predictors in order to perform model selection.
1254-1259
Ranalli, Maria Giovanna
aec65b36-08c7-467c-ae6e-77e8308cffd3
Petrella, Lea
44b14c98-47e9-4d16-aaad-839565652073
Pantalone, Francesco
c1b85bef-a71c-4851-9807-7776bc0b5ded
Schirripa Spagnolo, Francesco
2020
Ranalli, Maria Giovanna
aec65b36-08c7-467c-ae6e-77e8308cffd3
Petrella, Lea
44b14c98-47e9-4d16-aaad-839565652073
Pantalone, Francesco
c1b85bef-a71c-4851-9807-7776bc0b5ded
Schirripa Spagnolo, Francesco
Ranalli, Maria Giovanna, Petrella, Lea and Pantalone, Francesco
(2020)
M-quantile regression shrinkage and selection via the lasso.
Pollice, Alessio, Salvati, Nicola and Schirripa Spagnolo, Francesco
(eds.)
In Scientific meeting of the Italian Statistical Society - SIS 2020.
Pearson.
.
Record type:
Conference or Workshop Item
(Paper)
Abstract
Standard regression analysis investigates the average behavior of a response variable, y, given a vector of predictors, x. However, in some cases, the mean does not give a complete picture of a distribution. Therefore, quantile regression analyzes how the q-th quantile of the conditional distribution of y given x varies withx, and M-quantile regression generalizes this idea through the use of influence functions. When dealing with a large number of predictors, selection of a subset of themimproves the interpretability of the model. Towards this end, in this paper, we introduce M-quantile regression with lasso regularization. This allows us to investigate the extreme behavior of y conditional on x and to shrink the predictors in order to perform model selection.
Text
Pearson-SIS-2020-atti-convegno
- Version of Record
Restricted to Repository staff only
Request a copy
More information
Published date: 2020
Identifiers
Local EPrints ID: 479455
URI: http://eprints.soton.ac.uk/id/eprint/479455
PURE UUID: 5341774d-568a-475d-b60f-cf1a9b63ca63
Catalogue record
Date deposited: 24 Jul 2023 17:02
Last modified: 17 Mar 2024 04:10
Export record
Contributors
Author:
Maria Giovanna Ranalli
Author:
Lea Petrella
Author:
Francesco Pantalone
Editor:
Alessio Pollice
Editor:
Nicola Salvati
Editor:
Francesco Schirripa Spagnolo
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics