The Statistics Seminar speaker for Wednesday, October 30, 2019, is Haim Bar, an associate professor at the University of Connecticut. He received his Ph.D. in statistics at Cornell University in 2012. He received his M.Sc. in statistics in 2010 (Cornell University) and an M.Sc. in computer science in 2002 (Yale University). He received his bachelor degree in mathematics (Cum Laude) in 1993, at the Hebrew University in Jerusalem.
His professional interests include statistical modeling, shrinkage estimation, high throughput applications in biology (e.g., genomics, brain imaging), Bayesian statistics, variable selection, and machine learning.
From 1995 to 1997, he was with Motorola, Israel, as a computer programmer in the Wireless Access Systems Division. From 1997 until 2003 he worked for MicroPatent, LLC, where he held the position of Director of Software Development. In 2003 he moved to Ithaca, NY, and worked as a Principal Scientist at ATC-NY.
Talk: Quantile Regression Modelling via Location and Scale Mixtures of Normal Distributions
Abstract: We show that the estimating equations for quantile regression can be solved using a simple EM algorithm in which the M-step is computed via weighted least squares, with weights computed at the E-step as the expectation of independent generalized inverse-Gaussian variables. We compute the variance-covariance matrix for the quantile regression coefficients using a kernel density estimator that results in more stable standard errors than those produced by existing software. A natural modification of the EM algorithm that involves fitting a linear mixed model at the M-step extends the methodology to mixed effects quantile regression models. In this case, the fitting method can be justified as a generalized alternating minimization algorithm. Obtaining quantile regression estimates via the weighted least squares method enables model diagnostic techniques similar to the ones used in the linear regression setting.