Student sliced inverse regression

Florence Forbes 1 Alessandro Chiancone 1 Stéphane Girard 1
1 MISTIS - Modelling and Inference of Complex and Structured Stochastic Systems
Inria Grenoble - Rhône-Alpes, LJK - Laboratoire Jean Kuntzmann, INPG - Institut National Polytechnique de Grenoble
Abstract : Sliced Inverse Regression (SIR) has been extensively used to reduce the dimension of the predictor space before performing regression. SIR is originally a model free method but it has been shown to actually correspond to the maximum likelihood of an inverse regression model with Gaussian errors. This intrinsic Gaussianity of standard SIR may explain its high sensitivity to outliers as observed in a number of studies. To improve robustness, the inverse regression formulation of SIR is therefore extended to non-Gaussian errors with heavy-tailed distributions. Considering Student distributed errors it is shown that the inverse regression remains tractable via an Expectation-Maximization (EM) algorithm. The algorithm is outlined and tested in presence of outliers both on simulated and real data, showing improved results in comparison to a number of other existing approaches.
Type de document :
Communication dans un congrès
9th International Conference of the ERCIM WG on Computational and Methodological Statistics, Dec 2016, Seville, Spain
Liste complète des métadonnées

https://hal.archives-ouvertes.fr/hal-01415576
Contributeur : Stephane Girard <>
Soumis le : mardi 13 décembre 2016 - 12:07:14
Dernière modification le : mercredi 11 avril 2018 - 01:58:48

Identifiants

  • HAL Id : hal-01415576, version 1

Collections

Citation

Florence Forbes, Alessandro Chiancone, Stéphane Girard. Student sliced inverse regression. 9th International Conference of the ERCIM WG on Computational and Methodological Statistics, Dec 2016, Seville, Spain. 〈hal-01415576〉

Partager

Métriques

Consultations de la notice

367