Please note this is a repeat of the course that was held on 21-22nd March
Course Description
Spatial Interaction Models (SIMs) are statistical models used to predict origin-destination flows. They are widely applied within geography, planning, transportation and the social sciences to predict interactions or flows related to commuting, migration, access to services etc. They are also widely applied across the commercial sector for example to model flows of consumers between home and retail centres with broad applications in commercial decision making and policy evaluation.
This hands on course is designed to equip participants with the skills to build, calibrate and apply spatial interaction models suitable for addressing a broad range of research questions. We dont assume any prior knowledge of spatial interaction modelling and begin by building a SIM for modelling consumer flows between home and retail stores. This intuitively straightforward example is used to understand the model structure, key theoretical assumptions and the model building and calibration process. We work with this model to understand model disaggregation and we also use this example to highlight one of the major commercial applications of the SIM.
The second part of the course will explore how we can use SIMs to explain and predict flows of humans such as daily commuting flows or less frequent migration flows. We will explore how to build and calibrate a production-attraction constrained SIM using the powerful open source software package R. Techniques for fitting a SIM to existing flow data and using the model to estimate missing data or predict future flows will be explored. We will also be able to discuss your own potential applications of the SIM.
Objectives:
To introduce participants to the production-constrained and production-attraction constrained SIMs and their applications within geography, social sciences, planning and the commercial sector.
To enable participants to build and calibrate SIMs using Microsoft Excel and R, particularly within the application areas of modelling retail or migration flows.
To equip participants with the skills to apply their models to predict flows under various what if? scenarios and to estimate missing data.
To encourage participants to evaluate their modelling framework, to assess model performance and to identify opportunities for model enhancement.
Prerequisites:
Participants should have a good working knowledge of Microsoft Excel.
No prior knowledge of R is required as everything will be taught on the course, however some familiarity will be advantageous if you have no prior knowledge of programming at all. For absolute beginners, resources such as code schools R tutorialhttp://tryr.codeschool.com/ ;- or any of the resources recommended onhttps://www.rstudio.com/online-learning/#R will be good for gaining familiarity before the course.
Recommended Reading:
Birkin, M. and Clarke, G. P. 1991. Spatial interaction in geography. Geography Review,4(5), pp.16-21.
Wilson, A. G. 2010. Entropy in urban and regional modelling: retrospect and prospect.Geographical Analysis, 42(4), pp.364-394.
Dennett, A. 2012. Working Paper Series Paper 181 Estimating flows between geographical locations: get me started in spatial interaction modelling. London: Centre for Advanced Spatial Analysis, University College London.
Additional reading material will be recommended during the course.
Outline Programme (subject to minor change):
Tuesday 21st
From 10:30 Registration
11:00 – 13:00 – Introducing SIMs and building a basic disaggregate production-constrained SIM from scratch
13:00 – 13:45 - Lunch
13:45 – 15:30 - Model calibration and testing
15:30 – 15:45 - Coffee/tea
15:45 – 17:00 - Using the SIM to evaluate ‘What if?’ scenarios.
Wednesday 22nd
9:00 9:30 : Short introduction on SIMs for migration and commuting analysis
9:30 10:00: Setting up R and reading in data for the practical exercises
10:00 10:30 - Coffee/tea
10:30 12:00 Practical 4: Working with spatial interaction data in R and an introduction the ‘unconstrained’ spatial interaction model in R.
12:00 13:00 Lunch and opportunity to discuss your own potential applications of the SIM with course leaders.
13:00 14:45 Practical 5: Fitting Constrained Spatial Interaction Models in R – calibrating and interpreting parameters, generating more accurate flow estimates.
14:45 15:00 Feedback and concluding comments
