Simple linear regression is developed for one explanatory variable using the principle of least squares. The extension to two explanatory variables raises the issue of whether both variables are needed for a well-fitting model, or whether one is sufficient and, if so, which one.
These ideas are generalised to many explanatory variables (multiple regression), for which the necessary theory of linear models is developed in terms of vectors and matrices. Checking model adequacy is introduced, e.g. by examining plots of the residuals. Widening the class of models that can be considered by the use of dummy variables for qualitative explanatory variables to assess treatment effects.
The methods are implemented using a suitable software and students gain experience and advice through weekly worksheets.
One of the pre-requisites for MATH3012, MATH3013, MATH3014, MATH6021, MATH6025, MATH6027 and MATH6135
Pre-requisites: MATH1024 AND MATH2011