The following briefly outlines some key projects I've been involved in. See also
papers and publications and
talks, where you'll find more on these projects along with plenty of other work.
+
Neural Network Flood Prediction
-
This project built a system for flood prediction for the Environment Agency, using Artificial Neural Networks (ANN) - computer replications of how the brain works. This was, as far
as we know, the first system to do this in a real situation - previous systems were processed-based, and took, in some cases, days to run; the ANN system took seconds.
I was responsible, with Linda See, for building the
software.
+
Online Democracy
-
This project built a number of web-based mapping systems to improve public participation in planning processes associated with environmental risk, including nuclear waste disposal,
tree planting in the Yorkshire Dales, and urban renewal in Slaithwaite in the Pennines. The project built on experience with building online multi-layered maps in 1996, developing, as far as
we know, the first online mapping system which the public could upload information into. I was responsible for building the online mapping systems and analysing the data.
[More info]
Example papers:
- Evans, A.J., Kingston, R., and Carver, S. (2004) 'Democratic input into the nuclear waste disposal problem: the influence of geographical data on decision making examined through a web-based GIS' Journal of Geographical Systems, 6 (2), 117 - 132
- Kingston, R., Carver, S., Evans, A. and Turton, I. (2000) 'Web-based public participation geographical information systems: an aid to local environmental decision-making (pdf)' (see also abstract) Computers, Environments and Urban Systems, 24, 109-125.
+
Capturing Fuzzy/Vague Vernacular Geography
-
This project, with Tim Waters, looked to build online mapping software to capture "vernacular" geography - that is, vague or 'fuzzy'
places that people aren't exactly sure of: for example, "the shops" in the phrase "I'm going down the shops". We used the system to look at where people thought
was "high crime" and compare that to crime statistics. The idea is now the foundation for the joint Leeds-Lancaster-Manchester-USDA Forest Service Map-Me system, which
allows anyone to develop a similar website. The project re-invigorated the exploration of vernacular geography.
[More info]
Example papers:
+
Modelling Petrol Prices with Agent-Based Models
-
Alison Heppenstall's PhD, which I co-supervised.
A very early example of an agent-based model of a real economic system in real geographical space. It looked at the
effect of competition across different distances around a city. We validated it by looking at which stations had closed since the data was collected. The idea we put around of
differential petrol taxation for rural areas to fund public transport raises its head now and then. The project was, in some ways, too ahead of its time, as realistic (as opposed to abstract)
ABM papers in economics were almost impossible to get published in economics journals at the time - tell that to the kids of today.
Example papers:
- Heppenstall, A.J., Evans, A.J. and Birkin, M.H. (2003) 'A Hybrid Multi-Agent/Spatial Interaction Model System for Petrol Price Setting' Proceedings of the 7th International Conference on GeoComputation.
- Heppenstall, A.J., Evans, A.J. and Birkin, M.H. (2006) 'Using Hybrid Agent-Based Systems to Model Spatially-Influenced Retail Markets' Journal of Artificial Societies and Social Simulation, 9, 3.
+
Modelling Aphids as Agricultural Pests
-
Hazel Parry's PhD, funded by Defra. This looked at modelling realistic numbers of aphids within agricultural fields. Aphids have density-dependent movement, so understanding how the pest moves within fields involves
detailed individual-level modelling of millions of aphids, and their interactions - a hard computational task. We devised a nifty trick for turning this into a parallel model running on multiple machines, an algorithm
subsequently utilised by others with the same issues.
Example papers:
- Parry, H., Evans, A.J., and Morgan, D. (2006) 'Aphid population dynamics in agricultural landscapes: an agent-based simulation model' Ecological Modelling, Special Issue on Pattern and Processes of Dynamic Landscapes, 199 (4), 451-463.
- Parry, H. and Evans, A.J. (2008) 'A comparative analysis of parallel processing and super-individual methods for improving the computational performance of a large individual-based model' Ecological Modelling, 214 (2-4), 141-152
+
Glacial Mice
-
A chance encounter while clearing up an old field site in Iceland led us to look more closely at glacial "mice" - small tufts of moss growing on glacial surfaces. While we
didn't discover them, we were the first people to explain how they developed and evolved, and highlighted their importance in the ecology of glaciers, now an important research area.
Example papers:
- Porter, P.R, Evans, A.J., Hodson, A.J., Lowe, A.T. and Crabtree, M.D. (2008) 'Sediment-moss interactions on a temperate glacier: Falljokull, Iceland'. Annals of Glaciology, 48, 25-31
+
Crime Prediction
-
Initially starting with gravity models (where attractors like rich people pull in criminals), but then moving to Agent-Based Models (ABM; imagine SIMCity as an academic project),
a group of my PhD students have looked at modelling burglary. We built probably the first really detailed models of burglary for real cities, rather than
abstract cities, and I've also talked about some of the ethical issues around this kind of work, and how to avoid them. More recently, we
worked for the Home Office to assess predictive policing based on statistical treatments, and build open source versions of some of the major algorithms as
part of this. [More info][Predictive software]
Example papers:
- Malleson, N., See, L.M., Evans, A.J. and Heppenstall, A.J. (2010) 'Implementing comprehensive offender behaviour in a realistic agent-based model of burglary' Simulation, 0(0): 1-22
- Malleson N; Heppenstall A.J., See L.M.; Evans AJ; (2013) Using an agent-based crime simulation to predict the effects of urban regeneration on individual household burglary risk, E&P B, 40 doi:10.1068/b38057
+
Dynamic Data Assimilation in ABM
-
Traditionally, Agent-Based Modelling (a little like SIMCity, but for academic research) has been run in a "one shot" fashion - models are tweeked to match them to the real world, and then
run to predict the future. Instead, a group of us at Leeds have been pushing to use the same technologies used in weather and climate forecasting in social modelling. Dynamic Data Assimilation, as
it is known, takes streams of incoming data from the present moment, and uses them to tweek models as they continually run, giving better and better estimates of the future, and
resisting issues caused by models drifting away from reality.
Example papers:
- Evans, A.J. (2012) 'Uncertainty and Error'. In Heppenstall, A.J., Crooks, A.T., See, L.M., and Batty, M. (2011) 'Agent-Based Models of Geographical Systems'. Springer.
-
Ward, J., Evans, A.J. and Malleson, N.S. (2016) Dynamic calibration of agent based models using data assimilation. Royal Society Open Science.
