Dimension reduction strategies for modeling BI-clustered high dimensional data

Abstract (A three-stage framework is developed for fitting a mixture of regressions for high dimensional data. The method combines a hierarchical agglomerative grouping algorithm, regression-based clustering, and a sequential, group-wise sparse estimation called Layered Elastic Net selection (LENS)....

ver descrição completa

Detalhes bibliográficos
Autor principal: Supranes, Michael Van B.
Formato: Tese
Idioma:English
Publicado em: Quezon City University of the Philippines 2016.