Spatial clustering using a genetic algorithm

One of the fundamental steps in many types of spatial analysis is to aggregate point data into groups. Geographical information systems are often used to form spatial groups by aggregating small areal units into larger, contiguous areas that can be given a particular classification. A common problem with spatial clustering procedures is that the scale of areal unit chosen for the aggregation has a dramatic effect on the results of the classification. This effect is commonly known as the Modifiable Areal Unit Problem (MAUP). This paper presents a Genetic Algorithm (GA) that is used to cluster spatial data using a flexible representation of areal unit. By incoporating the areal unit into the clustering and classification procedure some of the problems of associated with MAUP can be overcome. The GA has been specifically designed to search for clusters of items that share characteristics to produce an accurate spatial classification.