Show/Hide Toolbars

MPCluster for Maptitude

Navigation: Running MPCluster

Writing the Clusters to New Layers

Scroll Prev Top Next More

MPCluster has the ability to draw the cluster centers and boundaries to Maptitude layers. These are created so that each cluster has its own color. Do this by setting the Write clusters to Layers check box in the Display Options box. Then enter a Layer Prefix. MPCluster will create two layers based on this prefix. The centers will be written to a point layer with the _CTR suffix, and the boundaries will be written to an area layer with the _BDR suffix.


MPCluster can also draw circles around clusters at the requested maximum diameter (or radius). These will be drawn on a layer with the _CIR prefix. Enable this functionality by checking the Draw Cluster Circles check box. This will only be enabled if you have specified a maximum cluster diameter or radius.


When you start processing, Maptitude will ask you for the dbd file prefix. This will be used to create the corresponding data files for the new layers. The resulting files will include suffixes of CTR, BDR, and optionally CIR (with no underline character).


You may enter a layer prefix for layers that already exist. In this situation, MPCluster will the existing files and layers, but it will overwrite the layer contents and corresponding themes.


The boundaries are drawn as simple perimeter shapes (i.e. no fill). K-Means boundaries are technically convex hulls around the component data points. Hierarchical clusters can be concave, so these are drawn as alpha shapes (aka concave hulls). There is a cut-off in how concave a cluster can be, so that result in boundary shapes that appear to slightly overlap adjacent cluster boundaries. Use the Data View output option to view precise cluster allocations in this scenario.


The boundary layer can also be labeled. Switch labels on using the Draw Labels check box. Maptitude typically draws the labels in the middle of the boundary polygon.


The centers are marked with a solid triangle.


Here is an example of the clusters marked in this way (data points have been hidden for clarity):




Typically you would also use this option with the data view output. The data view theme uses the same colors as the layers, so the data points, boundaries, and center points for each cluster.


Cluster boundary layers can also include overlay data fields. These list the sum or count of a series of selected data fields from a point layer. This point layer can be the input layer, or it can be another data layer.