620-372 Applied Statistical Analysis

One of 620-202 or 620-204. 620-371 is recommended.

This subject extends the theory of inference developed in 620-202 Statistics and demonstrates how it is applied in practice. In addition, a number of recently developed techniques for analysing data, which involve extensive computer computations, are considered. Students will develop an understanding of the principles of statistical inference and will learn to use a number of important specific techniques in applied statistics.

Topics covered include principles and fundamental results in estimation and hypothesis testing; including consistency, sufficiency, minimum variance unbiased estimation, likelihood methods and associated asymptotic theory, optimal tests and likelihood ratio tests. Application to a selection from the following specific areas is studied: logistic regression, survival analysis, time series, epidemic models and Markov chain models. Students also study selected topics from Bayesian methods, including the use of MCMC to derive posterior distributions; re-sampling methods; jack-knife and the bootstrap; use of the bootstrap for exploring the sampling distribution of an estimator; robust and non-parametric methods; density estimation methods; non-parametric regression; decision theory as applied to statistical inference; and further likelihood theory including the EM algorithm, REML.

Up to 50 pages of written assignments and a 3-hour end-of-semester written examination.