MA485 Applied Regression and Time Series Analysis

Regression analysis is much more than simply fitting a line through some data points. Regression is concerned with detecting and modeling functional relationships between the response variable distribution and one or more predictor variables. Regression methods, which include many common techniques such the t-test and ANOVA as special cases, have important applications in science, engineering, and business.

In addition to covering standard regression topics such as

• Classical multiple regression - matrix formulation and standard techniques



• Classical variable selection methods - stepwise and best-subsets regression



• Elementary time series models - autoregressive models



• Binary response models - logistic and probit models

• Ridge regression - a classical method for handling multicollinearity



• Lasso and Elastic Net - sparse regression methods for analyzing "large p, small n" data in which the number of variables p exceeds the number of observations n.

In order to provide recently developed methods, the freeware statistical programming environment R will be used in addition to Minitab. Prereqs: one of (MA 221, MA 212) plus one of (MA 223, MA382) or else permission from the instructor. If you have any questions, please e-mail me at inlow@rose-hulman.edu.