Special Topic in Statistics - Advanced Data Analysis and Statistical Consulting

Description

In most undergraduate courses, you are taught the theory behind a method and then given neat examples to which it can be applied and software to apply it. In reality, the most common question you will hear from a non-statistician is ’how do I analyse my data?' So you are the one who has to come up with the appropriate research question and choose the suitable method (and sometimes learn it quickly too). It is common in real world applications for the experiments to have not been well planned and for data to be missing, which will need to be taken it into account. The assumptions underlying the statistical model (e.g. homoscedasticity and normally distributed) often do not hold and you will have to know what to do. Finally, your fellow scientists, laymen and policymakers are all interested in different aspects of the research question and that is rarely the statistical significance of your ANOVA: you have to know how to communicate your results clearly, correctly and efficiently and how to defend your choices in data analysis and collection.

This course is about the reality of being an applied statistician. Besides covering the above points in class, individual statistical consulting session will provide you with hands-on experience.

Good knowledge of multivariate statistical methods, GLMs, and basic sampling theory expected. Working knowledge of R is recommended for forecasting methods. It provides extensive training in forecasting and modelling techniques such as smoothing, dynamic regressions, multivariate autoregressions, state space models, and neural networks with a wide range of applications.