Menu Content/Inhalt
Seminars Print
previous year previous month next month next year
See by year See by month See by week See Today Search Jump to month
Emilie Devijver, KULeuven Print
Thursday, 21 April 2016, 12:15 - 13:15

Emilie Devijver, KULeuven

Model-based clustering for high-dimensional regression data

Abstract: Finite mixture regression models are useful for modeling the relationship between response and predictors, arising from different subpopulations. We will work with a finite mixture regression model, with, in each cluster, a linear model.  We consider high-dimensional predictors and high-dimensional response. We propose a procedure to deal with this high-dimensional issue. The main idea is to construct a model collection, varying the sparsity. We conclude by selecting the best model using the slope heuristic. We use the Lasso estimator to take into account the coefficient sparsity, but refit estimators by the maximum likelihood estimator to avoid bias issue.  We derive an oracle inequality to justify the model selection step. 


Location: R42.2.113
Contact: Nancy De Munck, This e-mail address is being protected from spam bots, you need JavaScript enabled to view it