An introduction to spatio-temporal modelling of small-area data in R (few places remaining)
Date:
17/11/2020 - 18/11/2020
Organised by:
University of Glasgow
Presenter:
Professor Duncan Lee
Level:
Intermediate (some prior knowledge)
Contact:
Claire Spencer
NCRM Training and Capacity Building Co-ordinator
University of Manchester
claire.spencer@manchester.ac.uk
Venue: Online
Description:
Spatially referenced data come in many different forms, as observations could relate to either a single geographical location or a predefined geographical areal unit such as a local authority. Examples of the latter include the number of hospitalisations due to respiratory disease in Intermediate Zones in Scotland, or the percentage of children getting 5 or more GCSE grades at A - C in each local authority in England. These data are known as areal unit data, and are found in many fields in social science and beyond. There are many motivations for modelling such data, including identifying areas that are hotspots with high data values (e.g. identifying Intermediate Zones with a high risk of respiratory related hospitalisation), and estimating the impact of covariate factors on the data (e.g. what factors cause local authorities to exhibit poor educational attainment).
Researchers may have access to these data for a single time point, or may have repeated measurements over multiple years. The key challenge when modelling these data is spatial or spatio-temporal autocorrelation, whereby observations relating to nearby areal units or nearby time periods are likely to exhibit similar values. This autocorrelation violates the assumption of independence commonly made by linear regression models, making them an inappropriate tool for data analysis. As such, more complex statistical models are required that allow for such correlations.
This course is aimed at anybody with basic R and statistical knowledge that wants to learn about small-area spatial data modelling and gives a two-day introduction to modelling spatio-temporal areal unit data covering:
- Exploratory spatial data analysis (lecture and practical sessions)
- Spatial data modelling (lecture and practical sessions)
- Exploratory spatio-temporal data analysis (lecture and practical sessions)
- Spatio-temporal data modelling (lecture and practical sessions)
Course timetable - Day 1
- 10:00 - 10:30 - Session 1 - Introduction and icebreaker.
- 10:30 - 11:00 - Session 2 - Lecture on exploratory spatial data analysis.
- 11:00 - 11:30 - Virtual coffee break.
- 11:30 - 12:30 - Session 3 - Practical session on exploratory spatial data analysis.
- 12:30 - 13:30 - Virtual lunch break.
- 13:30 - 14:30 - Session 4 - Lecture on spatial data modelling and start of practical session.
- 14:30 - 15:00 - Virtual coffee break.
- 15:00 - 16:00 - Session 5 - Practical session on spatial data modelling.
Course timetable - Day 2
- 10:00 - 10:30 - Session 6 - Recap on day 1 and lecture on spatio-temporal exploratory data analysis.
- 10:30 - 11:15 - Session 7 - Practical session on spatio-temporal exploratory data analysis.
- 11:15 - 11:40 - Virtual coffee break.
- 11:40 - 12:30 - Session 8 - Lecture on spatio-temporal data modelling and start of practical session.
- 12:30 - 13:30 - Virtual lunch break.
- 13:30 - 14:30 - Session 9 - Practical session on spatio-temporal data modelling.
- 14:30 - 15:00 - Virtual coffee break.
- 15:00 - 16:00 - Session 10 - Question and answer session for participants to ask questions about their own data.
The course will use R and Rstudio for the analysis, and a basic knowledge of R is required (e.g. reading in data, basic plotting, etc). However, full R code will be provided to run the exemplar analyses. Participants will need the following software installed and access to Zoom.
R version 4.0.2
Rstudio version 1.3.1093
CARBayes package version 5.2
CARBayesST package version 3.1
ggplot2 package version 3.3.2
leaflet package version 2.0.3
maptools package version 1.0-2
RColorBrewer package version 1.1-2
rgdal package version 1.5-16
rgeos package version 0.5-5
sp package version 1.4-2
spdep package version 1.1-5
tidyr package version 1.1.2
Cost:
The fee per teaching day is:
• £30 per day for UK/EU registered students
• £60 per day for staff at UK/EU academic institutions, UK/EU Research Councils researchers, UK/EU public sector staff and staff at UK/EU registered charity organisations and recognised UK/EU research institutions.
• £100 per day for all other participants
All fees include event materials.
Cancellation Policy: Please be reminded that you will be charged the full registration fee if you cancel your place within 4 weeks before the training delivery date or you fail to attend.
If you are able to fill the place on the course you are cancelling then the cancellation charge will not apply.
Website and registration:
Region:
Scotland
Keywords:
Bayesian methods, Spatial Data Analysis, Area-based analysis, Spatio-temporal analysis, Exploratory data analysis
Related publications and presentations:
Bayesian methods
Spatial Data Analysis
Area-based analysis
Spatio-temporal analysis