Teaching
Teaching
This course was developed as a first, or introductory course in the statistical analysis of time-series. The course introduces models, methods and theory for univariate time-series. As this is billed as a statistics class it will be assumed that students have basic familiarity with linear algebra, differentiation, and basic manipulation of random variables. The course is not overly mathematical in its treatment and is focused on giving intuition behind different time-series models, the processes they can describe, and how to apply these models in a forecasting context.Â
This course forms part of the MSc Data-Science program as an optional choice in the second (Lent) semester. It focuses on practical aspects of machine learning from both a supervised and unsupervised perspective. Students will learn via a variety of R packages how to implement methods such as: Linear regression, Subset and Regularised Model Selection, Generalised Linear Models, Generalised Additive Models, Neural-Networks, Decision Trees, Random Forests, Bagged and Boosted predictive models, Principle (and Independent) Component analysis, k-means clustering, hierarchical clustering, Gaussian Mixture Models, Latent Class Analysis, Mixture of Regressions.
This is a masters level course in statistical inference from the likelihood perspective (MSc Statistics, MSc Data-Science - mathematical stream). We cover in moderate detail both practical and theoretical aspects of likelihood estimation: optimisation, testing, large sample, and asymptotic properties of maximum likelihood estimation. These will be explored via synthetic examples in the labs. We will also discuss estimation with dependent sampling, bootstrap methods for examining estimator variance.
This course is part of the MRes program for STOR-i students. It aims to develop basic research skills and prepare students for future careers and PhD study.
More details can be found here.