Examples with ClassificationResult gdsc.core.match.ClassificationResult used on opensource projects

Example 1 with ClassificationResult

use of gdsc.core.match.ClassificationResult in project GDSC-SMLM by aherbert.

the class FilterAnalysis method run.

private ClassificationResult run(Filter filter, List<MemoryPeakResults> resultsList) {
    ClassificationResult s = filter.score(resultsList);
    if (showResultsTable) {
        StringBuilder sb = new StringBuilder();
        sb.append(filter.getName()).append("\t");
        sb.append(s.getTP() + s.getFP()).append("\t");
        sb.append(s.getTP()).append("\t");
        sb.append(s.getFP()).append("\t");
        sb.append(s.getTN()).append("\t");
        sb.append(s.getFN()).append("\t");
        sb.append(s.getJaccard()).append("\t");
        sb.append(s.getPrecision()).append("\t");
        sb.append(s.getRecall()).append("\t");
        sb.append(s.getF1Score());
        if (isHeadless) {
            IJ.log(sb.toString());
        } else {
            resultsWindow.append(sb.toString());
        }
    }
    return s;
}

Example 2 with ClassificationResult

use of gdsc.core.match.ClassificationResult in project GDSC-SMLM by aherbert.

the class Filter method score.

/**
	 * Filter the results and return the performance score. Allows benchmarking the filter by marking the results as
	 * true or false.
	 * <p>
	 * Any input PeakResult with an original value that is not zero will be treated as a true result, all other results
	 * are false. The filter is run and the results are marked as true positive, false negative and false positive.
	 * 
	 * @param resultsList
	 *            a list of results to analyse
	 * @return the score
	 */
public ClassificationResult score(List<MemoryPeakResults> resultsList) {
    int tp = 0, fp = 0, tn = 0, fn = 0;
    for (MemoryPeakResults peakResults : resultsList) {
        setup(peakResults);
        for (PeakResult peak : peakResults.getResults()) {
            final boolean isTrue = peak.origValue != 0;
            boolean isPositive = accept(peak);
            if (isTrue) {
                if (isPositive)
                    // true positive
                    tp++;
                else
                    // false negative
                    fn++;
            } else {
                if (isPositive)
                    // false positive
                    fp++;
                else
                    // true negative
                    tn++;
            }
        }
        end();
    }
    return new ClassificationResult(tp, fp, tn, fn);
}

Example 3 with ClassificationResult

use of gdsc.core.match.ClassificationResult in project GDSC-SMLM by aherbert.

the class BenchmarkFilterAnalysis method score.

private ComplexFilterScore score(DirectFilter f, int i, int size, int[] combinations, boolean[] enable, int[] uniqueIds1, int uniqueIdCount1) {
    setupFractionScoreStore();
    // Score them
    long time = System.nanoTime();
    FilterScoreResult r = scoreFilter(f);
    time = System.nanoTime() - time;
    endFractionScoreStore();
    ClassificationResult r2 = null;
    if (uniqueIds1 != null) {
        // Build an overlap between the results created by this filter and the best filter.
        // Note that the overlap may be very low given the number of different ways we can generate
        // fit results (i.e. it is not as simple as just single-fitting on each candidate)
        // To do this we assign a unique ID to each possible result.
        // The two sets can be compared to produce a Precision, Recall, Jaccard, etc.
        // PreprocessedPeakResult will need an additional mutable Id field.
        // Sort by Id then iterate through both arrays concurrently counting the matching Ids.
        int[] uniqueIds2 = uniqueIds;
        int uniqueIdCount2 = uniqueIdCount;
        int tp = 0, fp = 0, fn = 0;
        // Compare Ids (must be sorted in ascending order)
        int i1 = 0, i2 = 0;
        while (i1 < uniqueIdCount1 && i2 < uniqueIdCount2) {
            final int result = uniqueIds1[i1] - uniqueIds2[i2];
            if (result > 0) {
                i2++;
                fp++;
            } else if (result < 0) {
                i1++;
                fn++;
            }
            i1++;
            i2++;
            tp++;
        }
        // Count the remaining ids
        fn += (uniqueIdCount1 - i1);
        fp += (uniqueIdCount2 - i2);
        r2 = new ClassificationResult(tp, fp, 0, fn);
    }
    // These scores are used when the same filter type so set allSameType to true
    return new ComplexFilterScore(r, true, i, time, r2, size, combinations, enable);
}

Example 4 with ClassificationResult

use of gdsc.core.match.ClassificationResult in project GDSC-SMLM by aherbert.

the class BenchmarkFilterAnalysis method createResult.

public StringBuilder createResult(DirectFilter filter, FractionClassificationResult r, String resultsPrefix2) {
    StringBuilder sb = new StringBuilder(resultsPrefix);
    sb.append(filter.getName()).append(resultsPrefix2).append(resultsPrefix3);
    int i = 0;
    // TODO - Fix the scores that we show since we no longer have TN results
    // We could set:
    // TN as any candidate that does not match a true result.
    // FN as any candidate that does match a true result (that is not matched by any fit result)
    // To do this properly would require that we store all the matches of candidates to the data.
    // These can then be totalled up given the candidates that have not been used to create a positive. 
    // Integer results
    final ClassificationResult r2 = (requireIntegerResults) ? createIntegerResult(r) : null;
    add(sb, r2.getTP(), i++);
    add(sb, r2.getFP(), i++);
    add(sb, r2.getFN(), i++);
    add(sb, r2.getPrecision(), i++);
    add(sb, r2.getRecall(), i++);
    add(sb, r2.getF1Score(), i++);
    add(sb, r2.getJaccard(), i++);
    addCount(sb, r.getTP(), i++);
    addCount(sb, r.getFP(), i++);
    addCount(sb, r.getFN(), i++);
    add(sb, r.getPrecision(), i++);
    add(sb, r.getRecall(), i++);
    add(sb, r.getF1Score(), i++);
    add(sb, r.getJaccard(), i++);
    return sb;
}

Example 5 with ClassificationResult

use of gdsc.core.match.ClassificationResult in project GDSC-SMLM by aherbert.

the class FilterAnalysis method calculateSensitivity.

private void calculateSensitivity(List<MemoryPeakResults> resultsList) {
    if (!calculateSensitivity)
        return;
    if (!bestFilter.isEmpty()) {
        IJ.showStatus("Calculating sensitivity ...");
        createSensitivityWindow();
        int currentIndex = 0;
        for (String type : bestFilterOrder) {
            IJ.showProgress(currentIndex++, bestFilter.size());
            Filter filter = bestFilter.get(type).filter;
            ClassificationResult s = filter.score(resultsList);
            String message = type + "\t\t\t" + Utils.rounded(s.getJaccard(), 4) + "\t\t" + Utils.rounded(s.getPrecision(), 4) + "\t\t" + Utils.rounded(s.getRecall(), 4);
            if (isHeadless) {
                IJ.log(message);
            } else {
                sensitivityWindow.append(message);
            }
            // List all the parameters that can be adjusted.
            final int parameters = filter.getNumberOfParameters();
            for (int index = 0; index < parameters; index++) {
                // For each parameter compute as upward + downward delta and get the average gradient
                Filter higher = filter.adjustParameter(index, delta);
                Filter lower = filter.adjustParameter(index, -delta);
                ClassificationResult sHigher = higher.score(resultsList);
                ClassificationResult sLower = lower.score(resultsList);
                StringBuilder sb = new StringBuilder();
                sb.append("\t").append(filter.getParameterName(index)).append("\t");
                sb.append(Utils.rounded(filter.getParameterValue(index), 4)).append("\t");
                double dx1 = higher.getParameterValue(index) - filter.getParameterValue(index);
                double dx2 = filter.getParameterValue(index) - lower.getParameterValue(index);
                addSensitivityScore(sb, s.getJaccard(), sHigher.getJaccard(), sLower.getJaccard(), dx1, dx2);
                addSensitivityScore(sb, s.getPrecision(), sHigher.getPrecision(), sLower.getPrecision(), dx1, dx2);
                addSensitivityScore(sb, s.getRecall(), sHigher.getRecall(), sLower.getRecall(), dx1, dx2);
                if (isHeadless) {
                    IJ.log(sb.toString());
                } else {
                    sensitivityWindow.append(sb.toString());
                }
            }
        }
        String message = "-=-=-=-";
        if (isHeadless) {
            IJ.log(message);
        } else {
            sensitivityWindow.append(message);
        }
        IJ.showProgress(1);
        IJ.showStatus("");
    }
}