Geostatistical models for public health applications
Module Aims:
This module covers the area of geostatistical modelling and its application to epidemiology. The overarching aim of the course is to allow the students to gain a deeper understanding of the general statistical modelling paradigm (formulation, estimation, diagnostic checking, prediction, scientific interpretation) in the specific context of geostatistical methods. The course will also aim to strengthen skills in statistical analysis in the R software environment through the use of specialised R packages (PrevMap, lme4).
Module Learning Outcomes:
By the end of this module students should be able to:
- Recognise the essential characteristics of a geostatistical problem
- Diagnose regression residuals for evidence of spatial correlation
- Formulate and fit stochastic models to geostatistical data that show evidence of spatial correlation
- Use these models to construct predictive maps of spatially varying phenomena
- Explain how spatial statistical methods can help to understand scientific processes that exhibit spatial variation.
Pre-requisites:
- Basic knowledge of R programming
- Good knowledge of generalised linear models
- Good knowledge of basic concepts of probability theory: density function; cumulative density function; quantile; marginal distribution; conditional distribution; moments of a random variable.
Teaching Strategy:
The module will consist of a mix of lectures and practical sessions. Examples drawn from tropical disease epidemiology and environmental epidemiology will be used to introduce new modelling approaches and real data-sets will be used for exercises in practical sessions.
Assessment:
Formative exercises during the module
Project assignment (100% of module grade)
Module Length: 4 days.