public static void executePCA(Preprocessing pr) throws Exception {
// Using PCA from EJML
// PrincipleComponentAnalysis pca = new PrincipleComponentAnalysis();
// pca.setup(Environment.hm_bookmark_tag.size(),
// Environment.hm_token_weight.size());
// BufferedReader br = new BufferedReader(new FileReader(
// new File("matrix")));
// String s = "";
// String[] arrStr;
// double[] nums = null;
// while ((s = br.readLine()) != null) {
// arrStr = s.split("\t");
// nums = new double[arrStr.length - 1];
// for (int i = 0; i < nums.length; i++) {
// nums[i] = Double.parseDouble(arrStr[i + 1]);
// }
// pca.addSample(nums);
// }
// pca.computeBasis(Environment.featureSize);
// System.out.println(pca.sampleToEigenSpace(nums));
PrincipleComponentAnalysis pca = new PrincipleComponentAnalysis();
// int bm_num = 68479;
// int tg_num = 11619;
pca.setup(Environment.hm_bookmark_tag.size(), Environment.hm_token_weight.size());
File mxIN = new File("Output4Stats/matrix");
File mxOUT = new File("Output4Stats/norm_matrix_ejml");
BufferedReader br = new BufferedReader(new FileReader(mxIN));
String s = "";
String[] arrStr;
double[] nums = null;
System.out.println("Start reading matrix...");
while ((s = br.readLine()) != null) {
arrStr = s.split("\t");
nums = new double[arrStr.length - 1];
for (int i = 0; i < nums.length; i++) {
nums[i] = Double.parseDouble(arrStr[i + 1]);
}
pca.addSample(nums);
}
br.close();
System.out.println("Start PCA...");
pca.computeBasis(Environment.featureSize);
// Write down to file
System.out.println("Start writing norm matrix...");
pr.writeNormMatrix(pca, mxIN, mxOUT);
}
