GROUP reads a similarity matrix produced by OVRLAP and organizes the most similar groups into hierarchical clusters. The cluster solution is saved and written as six vectors on the output Cluster Solution file. These vectors are described more completely in the file description documentation.
Program Invocation
The specific syntax for program invocation varies between operating systems. This document diplays the code for the AIX version.
opt: Program Options (none currently supported) ssss: The study number OMf: Input OVRLAP matrix file CSf: Output Cluster Solution file for the results
GROUP Control Card
GROUP: cc 01-05 The literal 'GROUP' study: cc 07-10 The study number for this run maxmin: cc 12 Grouping (max/min) option. col: cc 14 The collapse option.
The grouping option indicates how the next groups are to be selected for merging together. The default option (maximize) cause the next two most similar groups to be merged. Maximizing is normally desired when the input is a similarity matrix. If the input is a difference matrix, an 'M' should be put in this column so the least dissimilar groups are merged.
The collapse option indicates which method should be used by GROUP when a merge takes place. The default option (Average) will be performed if this column is blank. This means that when merging two groups, the corresponding columns and rows for the two groups will be averaged to form a single row and column which will represent the new group. If an 'M' is placed in this column the maximum or minimum values from the rows and columns will be used for the new group column and row, depending on which option was selected for maxmin.
Examples
@codap group - ssss om100 cs100 GROUP ssss M
This example will create a clustered solution using the maximum values from the grouped pairs when collapsing the rows and columns of the matrix instead of the standard "average" function.
@codap group - ssss om100 cs100 GROUP ssss
This example would perform just as in the previous example except that the default maxmin and collapse options would be used. Note that the same CSf00 file may be used for output from any OVRLAP Matrix file from the same data family. The Cluster Solution file, like the Task Factor file, continues to accept additional records as necessary. To display the current contents of any Cluster Solution file, run the CLUSXX program.