Typical workflow for kernel PCA - Typical workflow for kernel PCA - 5.3 English - 68552

AOCL API Guide (68552)
Document_ID
68552
Release_Date
2026-05-13
Version
5.3 English

The standard way of computing the kernel PCA using AOCL-DA is as follows.

  1. Initialize a aoclda.factorization.KernelPCA() object with options set in the class constructor.

  2. Compute the kernel PCA for your data matrix using aoclda.factorization.KernelPCA.fit().

  3. Perform further transformations as necessary using aoclda.factorization.KernelPCA.transform() or aoclda.factorization.KernelPCA.inverse_transform().

  4. Extract results from the aoclda.factorization.KernelPCA() object via its class attributes.

  1. Initialize a da_handle with da_handle_type da_handle_kernel_pca.

  2. Pass data to the handle using da_kernel_pca_set_data_?.

  3. Set the kernel type, number of components, and other parameters using da_options_set_? (see below).

  4. Compute the kernel PCA using da_kernel_pca_compute_?.

  5. Perform further transformations as necessary using da_kernel_pca_transform_? or da_kernel_pca_inverse_transform_?.

  6. Extract results using da_handle_get_result_?.