On the accuracy in high dimensional linear models and its application to genomic selection

Abstract : Genomic selection, a hot topic in genetics, consists in predicting breeding values of selection candidates, using a large number of genetic markers , due to the recent progress in molecular biology. One of the most popular method chosen by geneticists is Ridge regression. In this context, we focus on some predictive aspects of Ridge regression and present theoretical results regarding the accuracy criteria, i.e., the correlation between predicted value and true value. We show the influence of the singular values, the regularization parameter , and the projection of the signal on the space spanned by the rows of the design matrix. Asymptotic results, in a high dimensional framework, are also given, and we prove that the convergence to an optimal accuracy highly depends on a weighted projection of the signal on each subspace. We discuss also on how to improve the prediction. Last, illustrations on simulated and real data are proposed.
Document type :
Journal articles
Liste complète des métadonnées

Cited literature [29 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-01456310
Contributor : Charles-Elie Rabier <>
Submitted on : Sunday, March 11, 2018 - 11:20:04 AM
Last modification on : Friday, April 12, 2019 - 4:22:50 PM
Document(s) archivé(s) le : Tuesday, June 12, 2018 - 12:16:30 PM

File

RabierManginGruseaV2ForHal.pdf
Files produced by the author(s)

Identifiers

Citation

Charles-Elie Rabier, Brigitte Mangin, Simona Grusea. On the accuracy in high dimensional linear models and its application to genomic selection. Scandinavian Journal of Statistics, Wiley, 2019, 46 (1), pp.289-313. ⟨10.1111/sjos.12352⟩. ⟨hal-01456310v2⟩

Share

Metrics

Record views

1193

Files downloads

149