Regression Analysis

This course is closely linked with MATH3230.03W, Analysis of Variance, for which it is a prerequisite. Students will use the computer heavily in these courses, but no previous courses in computing are required.

MATH3330.03 will focus on linear models for the analysis of data on several predictor variables and a single response. The emphasis will be on practical applications rather than on the formal derivations of the models. The approach will require the use of matrix representations of the data, and the geometry of vector spaces, which will be reviewed in the course.

The first term (MATH3330.03) will cover the basic ideas of multiple regression, having reviewed in depth the elements of simple linear regression. The second term (MATH3230.03) will have a major focus on models with categorical variables as predictors (classical ANOVA, or Analysis Of Variance).

The nature of the course requires that students be involved on a constant basis with the material, and not fall behind.

The text is Bowerman/O'Connel,Linear Statistical Models, 2nd ed. (Duxbury).

The prerequisites are (i) a course in basic statistics with coverage of t- and F-statistics, as well as an introduction to simple linear regression (examples are MATH1131.03/1132.03, MATH2030.06, MATH2560.03/2570.03, PSYC2020.06); and (ii) some facility with linear algebra (including the idea of vectors), such as provided in MATH2021.03/2022.03 (formerly MATH2000.06), MATH2221.03/2222.03 (formerly MATH2220.06), or MATH1550.06, MATH1025.03, MATH1505.06.

The final grade may be based (in each term) on assignments, quizzes, one or more midterms, and a final examination which will be common to all sections.