CLUSTER can
read matrices written by a previous CLUSTER command or by PROXIMITIES,
or created by MATRIX DATA. When
the input matrix contains distances between variables, CLUSTER clusters all or a subset of the
variables.
Values for split-file variables should precede values
for ROWTYPE_. CASENO_ and the labeling variable (if present) should
come after ROWTYPE_ and before VARNAME_.
If CASENO_ is
of type string rather than numeric, it will be considered unavailable
and a warning is issued.
If CASENO_ appears
on a variable list, a syntax error results.
CLUSTER ignores
unrecognized ROWTYPE_ values.
When you are reading a matrix created with MATRIX DATA, you should supply a value label
for PROX of either SIMILARITY or DISSIMILARITY so that the matrix is correctly
identified. If you do not supply a label, CLUSTER assumes DISSIMILARITY. (See “Format of the Matrix Data File” below.)
The program reads variable names, variable and value
labels, and print and write formats from the dictionary of the matrix
data file.
MATRIX=IN cannot be specified unless an active
dataset has already been defined. To read an existing matrix data
file at the beginning of a session, use GET to retrieve the matrix file and then specify IN(*) on MATRIX.
The variable list on CLUSTER can be omitted when a matrix data file is used as input. By default,
all cases or variables in the matrix data file are used in the analysis.
Specify a variable list when you want to read in a subset of items
for analysis.