2023/24 Taught Postgraduate Module Catalogue

OMAT5205M Multivariate Methods

15 creditsClass Size: 150

Module manager: Dr Arief Gusnanto
Email: A.Gusnanto@leeds.ac.uk

Taught: 1 Jul to 31 Aug View Timetable

Year running 2023/24

N/A

Pre-requisites

 OMAT5101M Statistical Methods

Module replaces

N/A

This module is not approved as an Elective

Module summary

In big data with multiple variables, it is vital to discover pattern and infer valuable information from the data. This module introduces basic techniques from multivariate statistics, with the aim to discover, describe and exploit dependencies between variables in complex datasets.

Objectives

The module will equip students with techniques to deal with multi-dimensional data using the techniques stated in the learning outcomes. Students will understand the theory through practice by engaging with example data sets.

Learning outcomes
On completion of the module, the student should be able to:

1. discover and exploit dependency between variables;
2. reduce the dimension of a dataset with dependent components, and interpret the results;
3. identify clusters in a given data set;
4. visualise similarities between observations in lower dimensions.

Skills outcomes
Skills developed in this module include:

- evaluating the quality of data and selecting suitable methods of analysis.
- interpreting data and making decisions based on that interpretation.

Syllabus

- Introduction to multivariate analysis
- Statistical dependence, covariance matrix
- High dimensional problems, the "curse of dimensionality"
- Principal Component Analysis (PCA), dimension reduction
- Clustering, K-means method, distances between/within clusters
- Multidimensional Scaling (MDS)

Teaching methods

 Delivery type Number Length hours Student hours Lectures 5 1.00 5.00 Discussion forum 6 2.00 12.00 Seminar 1 1.50 1.50 Independent online learning hours 42.00 Private study hours 89.50 Total Contact hours 18.50 Total hours (100hr per 10 credits) 150.00

Opportunities for Formative Feedback

Students will have weekly formative assignments (e.g. quizzes, problem sheets or practical tasks) for each taught unit of the module and will be given model solutions with comments.

Methods of assessment

Coursework
 Assessment type Notes % of formal assessment In-course Assessment Students will be tested predominantly using e-assessment methods or MCQs. 20.00 Assignment The assignment will require students to complete a written report which may feature components of R code, R outputs, calculations and critical analysis of results. It is expected that the assignment will be completed in one week. 80.00 Total percentage (Assessment Coursework) 100.00

Students will resit by completing the Assignment (which covers all learning outcomes) six months after the delivery of the module.