The KERNEL statement specifies the type of kernel and any associated parameters to use during training. You can specify one of the following kernel-types and its associated kernel-parameters:
LINEAR
specifies that the kernel type is linear. When the kernel type is linear, regular PCA is performed.
POLYNOMIAL / DEGREE=number INTERCEPT=number
specifies that the kernel type is in the polynomial form. The degree of the polynomial is a positive integer value that is specified in the DEGREE= option. By default, DEGREE=3. The intercept is a numeric value that is specified in the INTERCEPT= option. By default, INTERCEPT=1.
RBF / BW=s
uses a radial basis function (also known as the Gaussian kernel function) during training. The kernel is defined as
where x and y are vectors and s is the bandwidth parameter. You can specify the bandwidth parameter in the form BW=s, where s must be a positive nonzero real number. The processing time usually increases as the value of s decreases. At very small values of s, the processing time can be very long.
By default, the value of BW = , where n is the number of rows in the training data. This choice of default value comes from the idea in Kwok and Tsang (2004). The default parameter is not guaranteed to produce good results. It is recommended that you tune this parameter by using some systematic approach, such as cross validation.