The aim of this module is to introduce the student to a number of major topics in modern statistical methods. The student will gain experience of applying these methods to real datasets and experience of reporting their findings/conclusions. Statistical software (R or equivalent) will be used.
OVERVIEW
A number of discrete topics, typically three, will be covered of which the following are indicative:
Classification
Tree based methods. Assessing classification accuracy; confusion matrix, specificity, sensitivity, OC curve & AUC. Compared to logistic regression methods.
Clustering
Multidimensional scaling; Hierarchical methods/k-means, distribution and density based clustering.
Factor analysis
Covariance and correlation matrices; principle components; factor analysis. Rotation of factor scores; How many factors to include.
Survival analysis
Censoring and incomplete data; Survivor and hazard functions; Life-table and KM methods; Log rank and Wilcoxon tests; PH models with regression structure.
Statistical network analysis
Manipulating relational data; Descriptive analysis of network graph characteristics; Exponential-family network models; Network block models.
Applied Bayesian data analysis
Prior predictive checking; Probabilistic generative models; Predictive posterior distributions; Hypothesis testing; Model selection.
A mix of live online classes and pre-recorded video lectures.
| Module Content & Assessment | |
|---|---|
| Assessment Breakdown | % |
| Other Assessment(s) | 100 |