This course extends previous work on the normal linear model and introduces the class of generalised linear models (GLM). For these classes of models the course covers model specification, estimation, inference, model building and model checking. It illustrates these concepts with practical applications across various domains.

40 lectures (2 each week in Weeks 1-10 of Semesters 1 and 2).

8 1-hour tutorials (4 each in Semesters 1 and 2).

7 2-hour labs (3 each in Semesters 1 and 2, plus 2h lab for assessment in Semester 1).

Regression Models (Level M) [STATS5025]

Regression Models (DD80) [STATS3016

Courses prescribed in the Honours or Master's programme to which the student has been admitted.

Two pieces of continuous assessment (one in each semester) - 20% (each piece worth 10%).

End of course examination lasting 120 minutes - 80%.

The aims of this course are:

■ to introduce the theory and methodology of normal linear models and generalised linear models.

■ to describe the main approaches for specifying, building, evaluating, and making inferences from these models.

■ to describe specific models within this wider class, including analysis of variance, multiple regression, logistic regression and models for count data;

■ to illustrate how to fit these models to real data sets from various application areas.

By the end of this course students will be able to:

■ Formulate statistically valid regression models for different kinds of data using distributions from the exponential family.

■ Describe the theory for how the models are fitted to data.

■ Build and fit models to real data using statistical software and interpret the results.

■ Perform statistical inference using the fitted models with reference to distributional theories

■ Compare a range of competing models using appropriate model comparison techniques.

■ Assess the validity of the assumptions underpinning these models for real data set