The maps shown here are extracted from an Arc/Info 6.1 application of the Besag and Newell (1991) spatial statistic (in an AML known as MAKIS1)to the analysis of urban land use clustering. The functionality of the Network module was used to provide a more sophisticated representation of notions of shop-front contiguity and proximity; results were tested for significance using Monte Carlo simulations.
The Besag and Newell method was operationalised in the following manner:
Step 1: Select a target site of specified land use. (e.g. footwear shops)
Step 2:Define a nearest neighbour level (e.g. NN = 4).
Step 3: Measure the distance to the nth (e.g. 4th) nearest neighbour of the same land use.
Step 4:Count the number of buildings within this distance.
Step 5:Randomly shuffle the land uses.
Step 6:Repeat Steps 4. and 5. to satisfy some pre-defined Monte Carlo simulation criterion.
Step 7:Select another target site from the specified land use.
Step 8:Repeat Steps 1. to 7. for all members of the specified land use.
Map 1. shows the location of the study site, narrowly leeward of the Oxfordshire/Northamptonshire boundary and since 1991 accessed by the umbilical M40 link between Oxford and Birmingham.
Map 2. gives a more detailed view of the relation between town and county.
Map 3. describes the town centre which was studied in toto for two reasons. Firstly, to detect clustering for a retail type that has not been too greatly agglomerated (e.g. 'services') a reasonable number of members for narrower categories is required. Secondly, the sense of a town as a totality, more than the sum of its parts, with emergent properties at that scale prompted study with respect for this inherent holism.
Map 4. shows visualisations of some early results obtained by running MAKIS1, where the reciprocal of the 4th nearest neighbour distance is used to determine the size of proportional circles for every member of the shop-type selected dark (blue) circles for clustering significant to 0.08 per cent with Monte Carlo simulation, lighter (yellow) circles for trials that failed this test.
