public static void outputTopWordsWithProbs(
double[][] topicWordDistr, ArrayList<String> vocab, int numTopWord, String filepath)
throws Exception {
BufferedWriter writer = IOUtils.getBufferedWriter(filepath);
for (int t = 0; t < topicWordDistr.length; t++) {
// sort words
double[] bs = topicWordDistr[t];
ArrayList<RankingItem<Integer>> rankedWords = new ArrayList<RankingItem<Integer>>();
for (int i = 0; i < bs.length; i++) {
rankedWords.add(new RankingItem<Integer>(i, bs[i]));
}
Collections.sort(rankedWords);
// output top words
writer.write("Topic " + (t + 1));
double cumm_prob = 0;
for (int i = 0; i < Math.min(numTopWord, vocab.size()); i++) {
cumm_prob += rankedWords.get(i).getPrimaryValue();
writer.write(
"\t"
+ vocab.get(rankedWords.get(i).getObject())
+ ", "
+ rankedWords.get(i).getPrimaryValue()
+ ", "
+ cumm_prob);
}
writer.write("\n");
}
writer.close();
}
public static void outputLibSVM(File outputFile, SparseVector[] features, int[][] labels) {
System.out.println("Outputing LIBSVM-formatted data to " + outputFile);
try {
BufferedWriter writer = IOUtils.getBufferedWriter(outputFile);
for (int ii = 0; ii < features.length; ii++) {
if (labels[ii].length == 0) {
continue;
}
// labels
for (int jj = 0; jj < labels[ii].length - 1; jj++) {
writer.write(labels[ii][jj] + ",");
}
writer.write(Integer.toString(labels[ii][labels[ii].length - 1]) + " ");
// features
for (int idx : features[ii].getSortedIndices()) {
double featureVal = features[ii].get(idx);
if (Math.abs(featureVal) < 10E-6) {
continue;
}
writer.write(" " + idx + ":" + features[ii].get(idx));
}
writer.write("\n");
}
writer.close();
} catch (Exception e) {
e.printStackTrace();
throw new RuntimeException(
"Exception while outputing " + "LIBSVM-formatted data to " + outputFile);
}
}
public static void outputDistribution(double[] distr, String filepath) throws Exception {
BufferedWriter writer = IOUtils.getBufferedWriter(filepath);
for (double d : distr) {
writer.write(d + " ");
}
writer.close();
}
/**
* Output top words for each topic with indices
*
* @param topicIndices List of topic indices
* @param topicWordDistr 2D array containing topical word distributions
* @param vocab List of tokens in the vocabulary
* @param numTopWord Number of top words to output
* @param filepath Path to the output file
*/
public static void outputTopWords(
ArrayList<Integer> topicIndices,
double[][] topicWordDistr,
ArrayList<String> vocab,
int numTopWord,
String filepath)
throws Exception {
BufferedWriter writer = IOUtils.getBufferedWriter(filepath);
for (int t = 0; t < topicWordDistr.length; t++) {
// sort words
double[] bs = topicWordDistr[t];
ArrayList<RankingItem<Integer>> rankedWords = new ArrayList<RankingItem<Integer>>();
for (int i = 0; i < bs.length; i++) {
rankedWords.add(new RankingItem<Integer>(i, bs[i]));
}
Collections.sort(rankedWords);
// output top words
writer.write("Topic " + topicIndices.get(t));
for (int i = 0; i < Math.min(numTopWord, vocab.size()); i++) {
writer.write("\t" + vocab.get(rankedWords.get(i).getObject()));
}
writer.write("\n");
}
writer.close();
}
public static void outputLogLikelihoods(ArrayList<Double> logLhoods, String filepath)
throws Exception {
BufferedWriter writer = IOUtils.getBufferedWriter(filepath);
for (int i = 0; i < logLhoods.size(); i++) {
writer.write(i + "\t" + logLhoods.get(i) + "\n");
}
writer.close();
}
public static void outputPerplexity(String outputFile, double perplexity) {
System.out.println("Outputing perplexity to " + outputFile);
try {
BufferedWriter writer = IOUtils.getBufferedWriter(outputFile);
writer.write(perplexity + "\n");
writer.close();
} catch (Exception e) {
e.printStackTrace();
throw new RuntimeException("Exception while outputing " + outputFile);
}
}
public static void outputLatentVariables(double[][] vars, String filepath) throws Exception {
BufferedWriter writer = IOUtils.getBufferedWriter(filepath);
StringBuilder line;
for (double[] var : vars) {
line = new StringBuilder();
for (double v : var) {
line.append(Double.toString(v)).append(" ");
}
writer.write(line.toString().trim() + "\n");
}
writer.close();
}
/** Output latent variable assignments */
public static void outputLatentVariableAssignment(int[][] var, String filepath) throws Exception {
StringBuilder outputLine;
BufferedWriter writer = IOUtils.getBufferedWriter(filepath);
for (int[] var_line : var) {
if (var_line.length == 0) {
writer.write("\n");
} else {
outputLine = new StringBuilder();
outputLine.append(Integer.toString(var_line.length)).append("\t");
for (int v : var_line) {
outputLine.append(Integer.toString(v)).append(" ");
}
writer.write(outputLine.toString().trim() + "\n");
}
}
writer.close();
}
public static void outputTopicCoherences(File outputFile, ArrayList<Double> topicCoherences) {
try {
BufferedWriter writer = IOUtils.getBufferedWriter(outputFile);
writer.write("Average-Coherence\t" + StatUtils.mean(topicCoherences) + "\n");
writer.write("Min-Coherence\t" + StatUtils.min(topicCoherences) + "\n");
writer.write("Max-Coherence\t" + StatUtils.max(topicCoherences) + "\n");
writer.write("Median-Coherence\t" + StatUtils.median(topicCoherences) + "\n");
for (Double topicCoherence : topicCoherences) {
writer.write(topicCoherence + "\n");
}
writer.close();
} catch (Exception e) {
e.printStackTrace();
throw new RuntimeException(
"Exception while outputing perplexity" + " results to " + outputFile);
}
}
