KPCA Procedure
PROC KPCA Statement
PROC KPCA <options>;
The PROC KPCA statement invokes the procedure. Table 1 summarizes the options available in the PROC KPCA statement.
Table 1: PROC KPCA Statement Options
| Option | Description |
|---|---|
| Input Data Table Option | |
| DATA= | Specifies the input data table |
| KPCA Options | |
| CENTER | Centers the input data |
| METHOD= | Specifies the method of performing KPCA |
| NTHREADS= | Specifies the number of threads to use on each computation node |
| ORDER | Specifies to order the observations before processing them |
| RANKTHRESHOLD= | Specifies the epsilon value to determine the rank of the kernel matrix |
| SCALE | Scales the input data |
| Output Table Options | |
| CENTROIDS= | Specifies the name of the centroids output table |
| EIGENVAL= | Specifies the name of the eigenvalue output table |
| EIGENVEC= | Specifies the name of the eigenvector output table |
You can specify the following options:
- CENTER
- CENTROIDS=libref.data-table
specifies the name of the output table for the centroids that are generated by clustering. This table is generated if METHOD=APPROXIMATE. libref.data-table is a two-level name, where libref refers to the library, and data-table specifies the name of the output data table. For more information about this two-level name, see the DATA= option and the section Using CAS Sessions and CAS Engine Librefs.
- DATA=libref.data-table
-
names the input data table for PROC KPCA to use. The default is the most recently created data table. libref.data-table is a two-level name, where
- libref
refers to a collection of information that is defined in the LIBNAME statement and includes the
library, which includes a path to the data, and a session identifier, which defaults to the active session but which can be explicitly defined in the LIBNAME statement. For more information about libref, see the section Using CAS Sessions and CAS Engine Librefs.- data-table
specifies the name of the input data table.
- EIGENVAL=libref.data-table
specifies the name of the output table that contains the eigenvalues. libref.data-table is a two-level name, where libref refers to the library, and data-table specifies the name of the output data table. For more information about this two-level name, see the DATA= option and the section Using CAS Sessions and CAS Engine Librefs.
- EIGENVEC=libref.data-table
specifies the name of the output table that contains the eigenvectors. libref.data-table is a two-level name, where libref refers to the library, and data-table specifies the name of the output data table. For more information about this two-level name, see the DATA= option and the section Using CAS Sessions and CAS Engine Librefs.
- METHOD=APPROXIMATE | EXACT
-
specifies the method to use in performing KPCA. You can specify the following values:
- APPROXIMATE
uses the low-rank approximation method.
- EXACT
uses the exact method.
By default, METHOD=APPROXIMATE.
- NTHREADS=number-of-threads
specifies the number of threads per computation node. The default value is the lesser of 16 and the number of threads available per computation node.
- ORDER
considers the order of the observations in the input data set on the basis of the
KPCA_ROWIDvariable (or theROWIDvariable if theKPCA_ROWIDvariable is absent) to generate consistent results when data are distributed on multiple grid nodes. For more information, see the section Low-Rank Approximation of KPCA Training Using the Nyström Method.- RANKTHRESHOLD=number
specifies the value of epsilon for determining the rank of the kernel matrix. The default number is
. - SCALE
scales the observations by the standard deviation of each column. If a constant variable exists (where all observations have the same value), the observations of this column are not scaled.