MPCluster is primarily designed to work with point layers ('datasets'), but it can also work with area layers. MPCluster will use the centroids for the shapes defined in an area layer. A centroid is simply a mathematical term for the center of the area shape. The centroids used are those that are reported by Maptitude. Note that the resulting cluster boundary shape (if requested) will also be defined by these centroids, so it will bisect the outer-lying area shapes, even though they are included in the cluster.
To use one of these datasets, simply selected the required dataset from the Input Dataset drop-down box on the main panel.
The distribution of these centroids will typically follow those of the parent geometry – e.g. a zipcode layer will have points distributed according to the regional zipcode geometry. This is of limited usefulness, so you would typically use these datasets with the Data Field functionality. These functions allow you to apply dataset fields to the clustering processing. For example, a data field can be summed and a constraining range applied (e.g. "each cluster must have between 1000 and 2000 customers"). Or a data field can be used as a weight which influences the center of the resulting cluster.