コード例 #1
0
  /**
   * Given the path to a file representing the text based serialization of a Linear Classifier,
   * reconstitutes and returns that LinearClassifier.
   *
   * <p>TODO: Leverage Index
   */
  public static LinearClassifier<String, String> loadFromFilename(String file) {
    try {
      BufferedReader in = IOUtils.readerFromString(file);

      // Format: read indices first, weights, then thresholds
      Index<String> labelIndex = HashIndex.loadFromReader(in);
      Index<String> featureIndex = HashIndex.loadFromReader(in);
      double[][] weights = new double[featureIndex.size()][labelIndex.size()];
      int currLine = 1;
      String line = in.readLine();
      while (line != null && line.length() > 0) {
        String[] tuples = line.split(LinearClassifier.TEXT_SERIALIZATION_DELIMITER);
        if (tuples.length != 3) {
          throw new Exception(
              "Error: incorrect number of tokens in weight specifier, line="
                  + currLine
                  + " in file "
                  + file);
        }
        currLine++;
        int feature = Integer.valueOf(tuples[0]);
        int label = Integer.valueOf(tuples[1]);
        double value = Double.valueOf(tuples[2]);
        weights[feature][label] = value;
        line = in.readLine();
      }

      // First line in thresholds is the number of thresholds
      int numThresholds = Integer.valueOf(in.readLine());
      double[] thresholds = new double[numThresholds];
      int curr = 0;
      while ((line = in.readLine()) != null) {
        double tval = Double.valueOf(line.trim());
        thresholds[curr++] = tval;
      }
      in.close();
      LinearClassifier<String, String> classifier =
          new LinearClassifier<String, String>(weights, featureIndex, labelIndex);
      return classifier;
    } catch (Exception e) {
      System.err.println("Error in LinearClassifierFactory, loading from file=" + file);
      e.printStackTrace();
      return null;
    }
  }