Cluster Properties

The following properties are available for the cluster tool:
Name
enables you to specify the name for this model.
  • Number of clusters specifies the number of clusters that are generated.
  • Seed specifies the seed value of the random number generator that is used during initial cluster assignments.
  • Initial assignment specifies the method that is used to create the initial cluster assignments. The available methods are:
    • Forgy specifies that k data points are selected at random to use as the centroids of the k clusters.
    • Random assigns observations to a cluster at random.
  • Visible roles determines how many effects are shown in the Cluster Matrix. Valid values are integers between 2 and 6, inclusive.
    When you specify a value n, the first n effects listed in the Variables table on the Roles tab are displayed. To change the effect pairs that are plotted in the Cluster Matrix, you can remove an effect from the analysis, and then immediately add it back in. The clustering results remain unchanged because you are using the same input data. However, the Variables table adds new effects to the bottom of the list.
  • Variable standardization transforms the effect variables so that they have a mean of zero and a standard deviation of 1. This property is enabled by default and affects the results displayed in the details table. The Cluster Matrix window and the Parallel Coordinates window display the original variables.
Parallel Coordinates
  • Number of bins specifies the number of bins used when generating the parallel coordinate polyline plots.
  • Maximum polylines specifies the maximum number of polylines generated by the parallel coordinate algorithm.
  • Visible roles determines how many effects are shown in the Parallel Coordinates plot. Valid values are integers between 2 and 10, inclusive.
enables you to display the cluster projection ellipses in the Cluster Matrix.
enables you to display the centroids in the Cluster Matrix.
Last updated: January 8, 2019