Search in sources :

Example 11 with NominalAttribute

use of smile.data.NominalAttribute in project smile by haifengl.

the class GAFeatureSelectionTest method testLearn.

/**
     * Test of learn method, of class GAFeatureSelection.
     */
@Test
public void testLearn() {
    System.out.println("learn");
    int size = 100;
    int generation = 20;
    ClassifierTrainer<double[]> trainer = new LDA.Trainer();
    ClassificationMeasure measure = new Accuracy();
    DelimitedTextParser parser = new DelimitedTextParser();
    parser.setResponseIndex(new NominalAttribute("class"), 0);
    try {
        AttributeDataset train = parser.parse("USPS Train", smile.data.parser.IOUtils.getTestDataFile("usps/zip.train"));
        AttributeDataset test = parser.parse("USPS Test", smile.data.parser.IOUtils.getTestDataFile("usps/zip.test"));
        double[][] x = train.toArray(new double[train.size()][]);
        int[] y = train.toArray(new int[train.size()]);
        double[][] testx = test.toArray(new double[test.size()][]);
        int[] testy = test.toArray(new int[test.size()]);
        GAFeatureSelection instance = new GAFeatureSelection();
        BitString[] result = instance.learn(size, generation, trainer, measure, x, y, testx, testy);
        for (BitString bits : result) {
            System.out.format("%.2f%% %d ", 100 * bits.fitness(), Math.sum(bits.bits()));
            for (int i = 0; i < x[0].length; i++) {
                System.out.print(bits.bits()[i] + " ");
            }
            System.out.println();
        }
        assertTrue(result[result.length - 1].fitness() > 0.88);
    } catch (Exception ex) {
        System.err.println(ex);
    }
}
Also used : DelimitedTextParser(smile.data.parser.DelimitedTextParser) AttributeDataset(smile.data.AttributeDataset) ClassificationMeasure(smile.validation.ClassificationMeasure) ClassifierTrainer(smile.classification.ClassifierTrainer) Accuracy(smile.validation.Accuracy) NominalAttribute(smile.data.NominalAttribute) BitString(smile.gap.BitString) Test(org.junit.Test)

Example 12 with NominalAttribute

use of smile.data.NominalAttribute in project smile by haifengl.

the class RandomForestTest method testUSPSNominal.

/**
     * Test of learn method, of class RandomForest.
     */
@Test
public void testUSPSNominal() {
    System.out.println("USPS nominal");
    DelimitedTextParser parser = new DelimitedTextParser();
    parser.setResponseIndex(new NominalAttribute("class"), 0);
    try {
        AttributeDataset train = parser.parse("USPS Train", smile.data.parser.IOUtils.getTestDataFile("usps/zip.train"));
        AttributeDataset test = parser.parse("USPS Test", smile.data.parser.IOUtils.getTestDataFile("usps/zip.test"));
        double[][] x = train.toArray(new double[train.size()][]);
        int[] y = train.toArray(new int[train.size()]);
        double[][] testx = test.toArray(new double[test.size()][]);
        int[] testy = test.toArray(new int[test.size()]);
        for (double[] xi : x) {
            for (int i = 0; i < xi.length; i++) {
                xi[i] = Math.round(255 * (xi[i] + 1) / 2);
            }
        }
        for (double[] xi : testx) {
            for (int i = 0; i < xi.length; i++) {
                xi[i] = Math.round(255 * (xi[i] + 1) / 2);
            }
        }
        Attribute[] attributes = new Attribute[256];
        String[] values = new String[attributes.length];
        for (int i = 0; i < attributes.length; i++) {
            values[i] = String.valueOf(i);
        }
        for (int i = 0; i < attributes.length; i++) {
            attributes[i] = new NominalAttribute("V" + i, values);
        }
        RandomForest forest = new RandomForest(attributes, x, y, 200);
        int error = 0;
        for (int i = 0; i < testx.length; i++) {
            if (forest.predict(testx[i]) != testy[i]) {
                error++;
            }
        }
        System.out.println(error);
        System.out.format("USPS OOB error rate = %.2f%%%n", 100.0 * forest.error());
        System.out.format("USPS error rate = %.2f%%%n", 100.0 * error / testx.length);
        double[] accuracy = forest.test(testx, testy);
        for (int i = 1; i <= accuracy.length; i++) {
            System.out.format("%d trees accuracy = %.2f%%%n", i, 100.0 * accuracy[i - 1]);
        }
        double[] importance = forest.importance();
        int[] index = QuickSort.sort(importance);
        for (int i = importance.length; i-- > 0; ) {
            System.out.format("%s importance is %.4f%n", train.attributes()[index[i]], importance[i]);
        }
        assertTrue(error <= 150);
    } catch (Exception ex) {
        System.err.println(ex);
    }
}
Also used : DelimitedTextParser(smile.data.parser.DelimitedTextParser) AttributeDataset(smile.data.AttributeDataset) Attribute(smile.data.Attribute) NominalAttribute(smile.data.NominalAttribute) NominalAttribute(smile.data.NominalAttribute) Test(org.junit.Test)

Example 13 with NominalAttribute

use of smile.data.NominalAttribute in project smile by haifengl.

the class BIRCHTest method testUSPS.

/**
     * Test of learn method, of class BIRCH.
     */
@Test
public void testUSPS() {
    System.out.println("USPS");
    DelimitedTextParser parser = new DelimitedTextParser();
    parser.setResponseIndex(new NominalAttribute("class"), 0);
    try {
        AttributeDataset train = parser.parse("USPS Train", smile.data.parser.IOUtils.getTestDataFile("usps/zip.train"));
        AttributeDataset test = parser.parse("USPS Test", smile.data.parser.IOUtils.getTestDataFile("usps/zip.test"));
        double[][] x = train.toArray(new double[train.size()][]);
        int[] y = train.toArray(new int[train.size()]);
        double[][] testx = test.toArray(new double[test.size()][]);
        int[] testy = test.toArray(new int[test.size()]);
        BIRCH birch = new BIRCH(x[0].length, 5, 16.0);
        for (int i = 0; i < 20; i++) {
            int[] index = Math.permutate(x.length);
            for (int j = 0; j < x.length; j++) {
                birch.add(x[index[j]]);
            }
        }
        birch.partition(10);
        AdjustedRandIndex ari = new AdjustedRandIndex();
        RandIndex rand = new RandIndex();
        int[] p = new int[x.length];
        for (int i = 0; i < x.length; i++) {
            p[i] = birch.predict(x[i]);
        }
        double r = rand.measure(y, p);
        double r2 = ari.measure(y, p);
        System.out.format("Training rand index = %.2f%%\tadjusted rand index = %.2f%%%n", 100.0 * r, 100.0 * r2);
        assertTrue(r > 0.65);
        assertTrue(r2 > 0.20);
        p = new int[testx.length];
        for (int i = 0; i < testx.length; i++) {
            p[i] = birch.predict(testx[i]);
        }
        r = rand.measure(testy, p);
        r2 = ari.measure(testy, p);
        System.out.format("Testing rand index = %.2f%%\tadjusted rand index = %.2f%%%n", 100.0 * r, 100.0 * r2);
        assertTrue(r > 0.65);
        assertTrue(r2 > 0.20);
    } catch (Exception ex) {
        System.err.println(ex);
    }
}
Also used : DelimitedTextParser(smile.data.parser.DelimitedTextParser) AttributeDataset(smile.data.AttributeDataset) NominalAttribute(smile.data.NominalAttribute) RandIndex(smile.validation.RandIndex) AdjustedRandIndex(smile.validation.AdjustedRandIndex) AdjustedRandIndex(smile.validation.AdjustedRandIndex) Test(org.junit.Test)

Example 14 with NominalAttribute

use of smile.data.NominalAttribute in project smile by haifengl.

the class NeuralNetworkTest method testUSPS.

/**
     * Test of learn method, of class NeuralNetwork.
     */
@Test
public void testUSPS() {
    System.out.println("USPS");
    DelimitedTextParser parser = new DelimitedTextParser();
    parser.setResponseIndex(new NominalAttribute("class"), 0);
    try {
        AttributeDataset train = parser.parse("USPS Train", smile.data.parser.IOUtils.getTestDataFile("usps/zip.train"));
        AttributeDataset test = parser.parse("USPS Test", smile.data.parser.IOUtils.getTestDataFile("usps/zip.test"));
        double[][] x = train.toArray(new double[train.size()][]);
        int[] y = train.toArray(new int[train.size()]);
        double[][] testx = test.toArray(new double[test.size()][]);
        int[] testy = test.toArray(new int[test.size()]);
        int p = x[0].length;
        double[] mu = Math.colMean(x);
        double[] sd = Math.colSd(x);
        for (int i = 0; i < x.length; i++) {
            for (int j = 0; j < p; j++) {
                x[i][j] = (x[i][j] - mu[j]) / sd[j];
            }
        }
        for (int i = 0; i < testx.length; i++) {
            for (int j = 0; j < p; j++) {
                testx[i][j] = (testx[i][j] - mu[j]) / sd[j];
            }
        }
        NeuralNetwork net = new NeuralNetwork(NeuralNetwork.ErrorFunction.CROSS_ENTROPY, NeuralNetwork.ActivationFunction.SOFTMAX, x[0].length, 40, Math.max(y) + 1);
        for (int j = 0; j < 30; j++) {
            net.learn(x, y);
        }
        int error = 0;
        for (int i = 0; i < testx.length; i++) {
            if (net.predict(testx[i]) != testy[i]) {
                error++;
            }
        }
        System.out.format("USPS error rate = %.2f%%%n", 100.0 * error / testx.length);
    } catch (Exception ex) {
        System.err.println(ex);
    }
}
Also used : DelimitedTextParser(smile.data.parser.DelimitedTextParser) AttributeDataset(smile.data.AttributeDataset) NominalAttribute(smile.data.NominalAttribute) Test(org.junit.Test)

Example 15 with NominalAttribute

use of smile.data.NominalAttribute in project smile by haifengl.

the class DecisionTreeTest method testUSPS.

/**
     * Test of learn method, of class DecisionTree.
     */
@Test
public void testUSPS() {
    System.out.println("USPS");
    DelimitedTextParser parser = new DelimitedTextParser();
    parser.setResponseIndex(new NominalAttribute("class"), 0);
    try {
        AttributeDataset train = parser.parse("USPS Train", smile.data.parser.IOUtils.getTestDataFile("usps/zip.train"));
        AttributeDataset test = parser.parse("USPS Test", smile.data.parser.IOUtils.getTestDataFile("usps/zip.test"));
        double[][] x = train.toArray(new double[train.size()][]);
        int[] y = train.toArray(new int[train.size()]);
        double[][] testx = test.toArray(new double[test.size()][]);
        int[] testy = test.toArray(new int[test.size()]);
        DecisionTree tree = new DecisionTree(x, y, 350, DecisionTree.SplitRule.ENTROPY);
        int error = 0;
        for (int i = 0; i < testx.length; i++) {
            if (tree.predict(testx[i]) != testy[i]) {
                error++;
            }
        }
        System.out.format("USPS error rate = %.2f%%%n", 100.0 * error / testx.length);
        assertEquals(328, error);
    } catch (Exception ex) {
        System.err.println(ex);
    }
}
Also used : DelimitedTextParser(smile.data.parser.DelimitedTextParser) AttributeDataset(smile.data.AttributeDataset) NominalAttribute(smile.data.NominalAttribute) Test(org.junit.Test)

Aggregations

NominalAttribute (smile.data.NominalAttribute)54 DelimitedTextParser (smile.data.parser.DelimitedTextParser)49 AttributeDataset (smile.data.AttributeDataset)48 Test (org.junit.Test)46 AdjustedRandIndex (smile.validation.AdjustedRandIndex)14 RandIndex (smile.validation.RandIndex)14 Attribute (smile.data.Attribute)12 ArrayList (java.util.ArrayList)7 EuclideanDistance (smile.math.distance.EuclideanDistance)5 IOException (java.io.IOException)4 BufferedReader (java.io.BufferedReader)3 ParseException (java.text.ParseException)3 LDA (smile.classification.LDA)3 InputStreamReader (java.io.InputStreamReader)2 DateAttribute (smile.data.DateAttribute)2 NumericAttribute (smile.data.NumericAttribute)2 StringAttribute (smile.data.StringAttribute)2 PlotCanvas (smile.plot.PlotCanvas)2 Accuracy (smile.validation.Accuracy)2 BorderLayout (java.awt.BorderLayout)1