KPCA Procedure

The "Model Information" table displays basic information about the parameters that are used in the procedure. This information includes the number of interval variables, the kernel type and kernel parameters, the threshold for nonzero eigenvalues, and the number of principal components that are generated.

The "k-means Clustering Information" table displays the parameters that are used in the k-means clustering method. This table is displayed only if METHOD=APPROXIMATE in the PROC KPCA statement. These parameters include the number of clusters, random seed, seed initialization method for k-means clustering, maximum number of iterations, and convergence criterion.

The "Number of Observations" table displays the number of observations that are read and used in the model. Observations that have missing values for at least one of the input variables are ignored.

The "Descriptive Statistics" table displays the mean and standard deviation of the input variables.

The "Task Timing" table displays the time in seconds and the percentage of time consumed by the various steps of the KPCA method. These steps include kernel matrix construction, k-means clustering, eigendecomposition, and scoring the training data.