Variable selection in robust heteroscedastic models with autoregressive covariance structures using EM-type algorithm

Date

2022-12-19

Author

Gökalp Yavuz, Fulya
GÜNEY, YEŞİM
ARSLAN, OLÇAY

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

104
views

0
downloads

The joint modeling of location and scatter matrix with multivariate t-distribution provides a valuable extension to the classical approach with normal distribution when the data set under consideration involves outliers or heavy tail outcomes. Variable selection is essential in these models since the covariance model has three models built into it, and the number of unknown parameters grows quadratically with the size of the matrix. The first purpose is to obtain the maximum likelihood estimates of the parameters and provide an expectation-maximization-type algorithm as an alternative to the Fisher scoring algorithm widely used for these models in the literature. Parameter estimation and variable selection are achieved simultaneously in a multivariate t joint location and scatter matrix model using shrinkage approaches. To assess the performance of the considered methods, we conducted a simulation study and real data analysis.

URI

http://www.cmstatistics.org/CMStatistics2022/docs/BoACFECMStatistics2022.pdf?20221205001440
https://hdl.handle.net/11511/101402

Conference Name

15th International Conference of the ERCIM WG on Computational and Methodological Statistics (CMStatistics 2022)

Collections

Department of Statistics, Conference / Seminar

Suggestions

OpenMETU
Core

Variable Selection with Shrinkage Methods in Elliptical Linear Mixed Models Gökalp Yavuz, Fulya (null; 2017-01-27) Linear Mixed Models (LMM) provide a common and convenient framework for the analysis of longitudinal and clustered data by incorporating random effects. The main assumption of classical LMM is having normally distributed random effects and error terms. However, there are several situations in which we need to use more robust distributions rather than (multivariate) normal to handle heavy tailed data or outliers. In this study, we aim to do variable selection in LMM with elliptically distributed random effec...
Classification models based on Tanaka's fuzzy linear regression approach: The case of customer satisfaction modeling ŞİKKELİ, GİZEM; KÖKSAL, GÜLSER; Batmaz, İnci; TÜRKER BAYRAK, ÖZLEM (IOS Press, 2010-01-01) Fuzzy linear regression (FLR) approaches are widely used for modeling relations between variables that involve human judgments, qualitative and imprecise data. Tanaka's FLR analysis is the first one developed and widely used for this purpose. However, this method is not appropriate for classification problems, because it can only handle continuous type dependent variables rather than categorical. In this study, we propose three alternative approaches for building classification models, for a customer satisf...
Hypothesis testing in one-way classification AR(1) model with Student’s t innovations: An application to a real life data Yıldırım, Özgecan; Yozgatlıgil, Ceylan; Şenoğlu, Birdal (null; 2017-05-26) In this study, we estimate the model parameters in one-way classification AR (1) model when the distribution of the error terms is independently and identically distributed (iid) Student’s t. Maximum likelihood (ML) methodology is used in the estimation procedure. We also introduce the F statistic based on the ML estimators of the parameters for testing the equality of the treatment means. See also Yıldırım (2017) (M.S. Thesis, METU, Ankara, Continue) and Şenoğlu and Bayrak (2016) (Linear Contrasts in one-w...
Evidence Optimization for Consequently Generated Models Strijov, Vadim; Krymova, Katya; Weber, Gerhard Wilhelm (2010-12-04) To construct as adequate regression model one has to fulfill the set of measured features with their generated derivatives. Often the number of these features exceeds the number of the samples in the data set. After a feature generation process the problem of feature selection from a set of highly correlated features arises. The proposed algorithm uses an evidence maximization procedure to select a model as a subset of generated features. During the selection process it rejects multicollinear features. A pr...
Gibbs Sampling in Inference of Copula Gaussian Graphical Model Adapted to Biological Networks Purutçuoğlu Gazi, Vilda (2017-09-01) Markov chain Monte Carlo methods (MCMC) are iterative algorithms that are used in many Bayesian simulation studies, where the inference cannot be easily obtained directly through the defined model. Reversible jump MCMC methods belong to a special type of MCMC methods, in which the dimension of parameters can change in each iteration. In this study, we suggest Gibbs sampling in place of RJMCMC, to decrease the computational demand of the calculation of high dimensional systems. We evaluate the performance of...

Citation Formats

F. Gökalp Yavuz, Y. GÜNEY, and O. ARSLAN, “Variable selection in robust heteroscedastic models with autoregressive covariance structures using EM-type algorithm,” presented at the 15th International Conference of the ERCIM WG on Computational and Methodological Statistics (CMStatistics 2022), Londrina, Brezilya, 2022, Accessed: 00, 2023. [Online]. Available: http://www.cmstatistics.org/CMStatistics2022/docs/BoACFECMStatistics2022.pdf?20221205001440.