A spatial data science approach to model-based clustering and semi-supervised variable selection

optional courses

The classes will be conducted by a visiting scholar dr Nema Dean. The course will be coordinated by an onsite lecturer – mgr Maria Kubara, while the whole class material will be delivered by the visiting professor.

The classes will be conducted by a visiting scholar dr Nema Dean. The course will be coordinated by an onsite lecturer – mgr Maria Kubara, while the whole class material will be delivered by the visiting professor.

The course will be taught in an intensive workshop setting over the course of two weeks in October (daily meetings). The students are asked to bring their own laptops with R v.3.3.0+ and RStudio Desktop installed in order to take active part in the practical live code exercises discussed during the class.

---------

Instructor:

Dr Nema Dean

School of Mathematics & Statistics

University of Glasgow, United Kingdom

Nema.Dean@glasgow.ac.uk

The list of course topics:

- A theoretical and practical introduction to non-parametric and parametric clustering using R

- Cluster comparison metrics and recent extensions

- A quick introduction to Bayesian CAR models for spatial modelling and their use in boundary detection (using the CARBayes R package)

- Use of clustering in spatial models

In this course, you will explore the fundamentals of clustering and spatial modeling using R, a versatile programming language widely used in data analysis. The topics covered include both non-parametric and parametric clustering, allowing you to gain insights into organizing and understanding complex datasets. You will learn about cluster comparison metrics, along with their recent extensions, enabling you to evaluate and compare different clustering methods effectively. Additionally, the course will introduce you to Bayesian Conditional Autoregressive (CAR) models, which are essential in spatial modeling and boundary detection. By combining these techniques, you will be equipped with valuable skills to analyze and interpret spatial data, making informed decisions and solving real-world problems across various domains.

- own materials

Literature:

- Frontiers in residential segregation: understanding neighbourhood boundaries and their impacts

N Dean, G Dong, A Piekut, G Pryce (2019)

Tijdschrift voor economische en sociale geografie 110 (3), 271-288

- sARI: a soft agreement measure for class partitions incorporating assignment probabilities

A Flynt, N Dean, R Nugent (2019)

Advances in Data Analysis and Classification 13, 303-323

- Spatial clustering of average risks and risk trends in Bayesian disease mapping

C Anderson, D Lee, N Dean (2017)

Biometrical Journal 59 (1), 41-56

- A Survey of Popular R Packages for Cluster Analysis

A Flynt, N Dean (2016)

Journal of Educational and Behavioral Statistics 41 (2), 205-225

- Identifying clusters in Bayesian disease mapping

C Anderson, D Lee, N Dean (2014)

Biostatistics 15 (3), 457-469

After this course the student:

• Gain a solid understanding of clustering techniques in data analysis using R.

• Be proficient in both non-parametric and parametric clustering methods.

• Understand cluster comparison metrics and their recent extensions for effective evaluation.

• Be introduced to Bayesian CAR models for spatial modeling and boundary detection using the CARBayes R package.

• Acquire essential skills to analyze and interpret spatial data in various applications.

• Have the ability to make informed decisions and solve real-world problems by applying clustering and spatial modeling.

K_U02, K_U05

The final grade will be based on the exam result.