Tilbake til søkeresultatene

FRINATEK-Fri prosj.st. mat.,naturv.,tek

Adaptive penalisation for p>>n sparse problems

Tildelt: kr 3,1 mill.

High dimensional regression modelling, where the number of covariates p is much larger than the number of samples n, is one of the most active research areas in statistics, as such situations frequently appear in many different areas, including molecular biology. When p>>n, classical statistics fails, and new penalised methods are available which exploit sparsity, as for example the lasso. In terms of variable selection, these methods are not stable and in certain contexts, suffer from lack of robustness and overfitting. In this project we will develop new more robust variants of adaptive penalised methods. Our approach is to strengthen and guide the variable selection procedure in two directions: (i) integrating into the analysis additonal informative so urces of data, tilting the penalisation to be coherent to both data sets, and (ii) leaving the linear models and developing semi-parametric approaches extending our method of parametrically guided non-parametric regression to the p>>n setting. Overfitting typically appears in multi-step procedures, where covariates are pre-selected using fast and simple (univariate) methods first, and penalisation is then performed on a reduced number of variables. We will develop new methods to correct for pre-selection bias. The new methods will be tested on high throughput genomic data like gene expressions, copy numbers, SNPs, etc. with collaborators in cancer genomics. Integration of data is one of the key challenges in system biology, and this project will propose new statistical tools for data integration. We hope that our approach will turn useful in identifying new genes and gene-environment interactions that play a role in the progression and therapy of ovarian and cervix cancer.

Publikasjoner hentet fra Cristin

Ingen publikasjoner funnet

Ingen publikasjoner funnet

Ingen publikasjoner funnet

Ingen publikasjoner funnet

Budsjettformål:

FRINATEK-Fri prosj.st. mat.,naturv.,tek