com.expleague.ml.loss.multiclass.util.MultilabelExampleTableOutput Maven / Gradle / Ivy

Go to download

Show more of this group Show more artifacts with this name
Show all versions of jmll

Various ML methods implemented by myself and my students

There is a newer version: 1.4.9

package com.expleague.ml.loss.multiclass.util;

import com.expleague.commons.math.vectors.Mx;
import com.expleague.ml.data.tools.Pool;

/**
 * Created by irlab on 25.06.2015.
 */
public class MultilabelExampleTableOutput {
  private final Mx scores;
  private final Mx targets;
  private final Pool pool;
  private final String name;

  public MultilabelExampleTableOutput(final Mx scores, final Mx targets, final Pool pool, final String name) {
    this.scores = scores;
    this.targets = targets;
    this.pool = pool;
    this.name = name;
  }

  public String toExampleTableMatrix() {
    final StringBuilder out = new StringBuilder();
    out.append(name);

    out.append("example");
    for (int i = 0; i < targets.columns(); i++)
      out.append("\tclass ").append(i).append(" score");
    for (int i = 0; i < targets.columns(); i++)
      out.append("\tclass ").append(i).append(" target");
    out.append('\n');

    for (int example = 0; example < targets.rows(); example++) {
      out.append(pool.data().at(example).id());
      final double[] tableRow = new double[targets.columns() * 2];
      for (int classId = 0; classId < targets.columns(); classId++) {
        final double score = 1. / (1. + Math.exp(-scores.get(example, classId)));
        out.append('\t').append(score);
      }
      for (int classId = 0; classId < targets.columns(); classId++) {
        out.append('\t').append(targets.get(example, classId));
      }
      out.append('\n');
    }

    return out.toString();
  }
}