| Biography | |
|---|---|
 Dr. Mahdi Roozbeh Semnan University, Iran |
|
| Title: Improved robust semiparametric regression analysis to combat outliers and multicollinearity problems in high-dimensional data to the comparative case studies with support vector machines | |
| Abstract: In classical regression analysis, the ordinary least-squares approach is the best method if the essential assumptions such as normality and independency to the error terms as well as a little or no multicollinearity in the explanatory variables are met. Nowadays, high-dimensional (wide) data in which the number of observations is smaller than the number of parameters as well as multicollinear data, appear in many practical applications such as biosciences, social networks, psychological researches, recommendation systems and so on. In these situations, the least-squares estimator is not applicable or the results can be misleading. Also, outliers violate the assumption of normally distributed residuals in the least-squares regression and robust modern approach should be used to analyse the contaminated data with outliers in high-dimensional case. Here, using the penalized scheme (such as LASSO) with the aim of obtaining a subset of effective explanatory variables that predict the response as the best, we develop a robust penalized mixed-integer nonlinear programming approach in high-dimension regression analysis in the presence of outliers and ill-conditioning problems. To complete the procedure, the generalized cross validation (GCV), which creates a balance between the precision of the estimators and the biasness caused by the penalization, is used to choose optimum value of the shrinkage parameter. It behaves like an improved estimator of risk and can be used when the number of explanatory variables is larger than the sample size in high-dimensional problems. Finally, performance of the proposed estimator is compared with the classical ones via a Monte-Carlo simulation and a real data example. | |
| Biography: My name is Dr. Mahdi Roozbeh and I am an Associate Professor of Mathematics, Statistics, and Computer Science Department since 2018 at Semnan University. I graduated with a Ph.D. in Statistics from the Ferdowsi University of Mashhad in 2011, and I am an expert in regression modeling and analysis of high-dimensional data. I am winner of ISI (International Statistical Institute) Jan Tinbergen Award (International Statistical Study Fund Foundation for the young statisticians and the prize for excellence in statistics is awarded every two years) in 2011 (which hold every 2 years in the world) for the best paper award during the World Statistical Congress in Dublin, awards for Research Excellence at Semnan University in 2015; 2017; 2018, Prof Behboodian Award and selected as the second Iranian young researcher in Statistics by Iranian Statistical Institute (2018), Elected for World Bank Trust Fund in the 62nd ISI WSC in Kuala Lumpur, 2019. My research focuses on high-dimensional deep learning, robust estimation and semiparametric regression modeling, and I am an expert in machine learning and published more than 40 papers indexed by ISI base yet. | |
