LogisticRegression.java

package com.github.lwhite1.tablesaw.api.ml.classification;

import com.github.lwhite1.tablesaw.api.BooleanColumn;
import com.github.lwhite1.tablesaw.api.CategoryColumn;
import com.github.lwhite1.tablesaw.api.IntColumn;
import com.github.lwhite1.tablesaw.api.NumericColumn;
import com.github.lwhite1.tablesaw.api.ShortColumn;
import com.github.lwhite1.tablesaw.util.DoubleArrays;
import com.google.common.base.Preconditions;

import java.util.SortedSet;
import java.util.TreeSet;

/**
 *
 */
public class LogisticRegression extends AbstractClassifier {

    private final smile.classification.LogisticRegression classifierModel;

    public static LogisticRegression learn(ShortColumn labels, NumericColumn... predictors) {
        smile.classification.LogisticRegression classifierModel =
                new smile.classification.LogisticRegression(DoubleArrays.to2dArray(predictors), labels.toIntArray());
        return new LogisticRegression(classifierModel);
    }

    public static LogisticRegression learn(IntColumn labels, NumericColumn... predictors) {
        smile.classification.LogisticRegression classifierModel =
                new smile.classification.LogisticRegression(DoubleArrays.to2dArray(predictors), labels.data()
                        .toIntArray());
        return new LogisticRegression(classifierModel);
    }

    public static LogisticRegression learn(BooleanColumn labels, NumericColumn... predictors) {
        smile.classification.LogisticRegression classifierModel =
                new smile.classification.LogisticRegression(DoubleArrays.to2dArray(predictors), labels.toIntArray());
        return new LogisticRegression(classifierModel);
    }

    public static LogisticRegression learn(CategoryColumn labels, NumericColumn... predictors) {
        smile.classification.LogisticRegression classifierModel =
                new smile.classification.LogisticRegression(DoubleArrays.to2dArray(predictors), labels.data()
                        .toIntArray());
        return new LogisticRegression(classifierModel);
    }

    public static LogisticRegression learn(ShortColumn labels, double lambda, NumericColumn... predictors) {
        smile.classification.LogisticRegression classifierModel =
                new smile.classification.LogisticRegression(DoubleArrays.to2dArray(predictors), labels.toIntArray(),
                        lambda);
        return new LogisticRegression(classifierModel);
    }

    public static LogisticRegression learn(IntColumn labels, double lambda, NumericColumn... predictors) {
        smile.classification.LogisticRegression classifierModel =
                new smile.classification.LogisticRegression(DoubleArrays.to2dArray(predictors), labels.data()
                        .toIntArray(), lambda);
        return new LogisticRegression(classifierModel);
    }

    public static LogisticRegression learn(BooleanColumn labels, double lambda, NumericColumn... predictors) {
        smile.classification.LogisticRegression classifierModel =
                new smile.classification.LogisticRegression(DoubleArrays.to2dArray(predictors), labels.toIntArray(),
                        lambda);
        return new LogisticRegression(classifierModel);
    }

    public static LogisticRegression learn(CategoryColumn labels, double lambda, NumericColumn... predictors) {
        smile.classification.LogisticRegression classifierModel =
                new smile.classification.LogisticRegression(DoubleArrays.to2dArray(predictors), labels.data()
                        .toIntArray(), lambda);

        return new LogisticRegression(classifierModel);
    }

    public static LogisticRegression learn(ShortColumn labels,
                                           double lambda,
                                           double tolerance,
                                           int maxIters,
                                           NumericColumn... predictors) {

        smile.classification.LogisticRegression classifierModel =
                new smile.classification.LogisticRegression(
                        DoubleArrays.to2dArray(predictors),
                        labels.toIntArray(),
                        lambda,
                        tolerance,
                        maxIters);
        return new LogisticRegression(classifierModel);
    }

    public static LogisticRegression learn(IntColumn labels,
                                           double lambda,
                                           double tolerance,
                                           int maxIters,
                                           NumericColumn... predictors) {

        smile.classification.LogisticRegression classifierModel =
                new smile.classification.LogisticRegression(
                        DoubleArrays.to2dArray(predictors),
                        labels.data().toIntArray(),
                        lambda,
                        tolerance,
                        maxIters);
        return new LogisticRegression(classifierModel);
    }

    public static LogisticRegression learn(BooleanColumn labels,
                                           double lambda,
                                           double tolerance,
                                           int maxIters,
                                           NumericColumn... predictors) {

        smile.classification.LogisticRegression classifierModel =
                new smile.classification.LogisticRegression(
                        DoubleArrays.to2dArray(predictors),
                        labels.toIntArray(),
                        lambda,
                        tolerance,
                        maxIters);
        return new LogisticRegression(classifierModel);
    }


    public static LogisticRegression learn(CategoryColumn labels,
                                           double lambda,
                                           double tolerance,
                                           int maxIters,
                                           NumericColumn... predictors) {

        smile.classification.LogisticRegression classifierModel =
                new smile.classification.LogisticRegression(
                        DoubleArrays.to2dArray(predictors),
                        labels.data().toIntArray(),
                        lambda,
                        tolerance,
                        maxIters
                );

        return new LogisticRegression(classifierModel);
    }

    private LogisticRegression(smile.classification.LogisticRegression classifierModel) {
        this.classifierModel = classifierModel;
    }

    public int predict(double[] data) {
        return classifierModel.predict(data);
    }

    public ConfusionMatrix predictMatrix(ShortColumn labels, NumericColumn... predictors) {
        Preconditions.checkArgument(predictors.length > 0);

        SortedSet<Object> labelSet = new TreeSet<>(labels.asSet());
        ConfusionMatrix confusion = new StandardConfusionMatrix(labelSet);

        populateMatrix(labels.toIntArray(), confusion, predictors);
        return confusion;
    }

    public ConfusionMatrix predictMatrix(IntColumn labels, NumericColumn... predictors) {
        Preconditions.checkArgument(predictors.length > 0);

        SortedSet<Object> labelSet = new TreeSet<>(labels.asSet());
        ConfusionMatrix confusion = new StandardConfusionMatrix(labelSet);

        populateMatrix(labels.data().toIntArray(), confusion, predictors);
        return confusion;
    }

    public ConfusionMatrix predictMatrix(CategoryColumn labels, NumericColumn... predictors) {
        Preconditions.checkArgument(predictors.length > 0);

        SortedSet<String> labelSet = new TreeSet<>(labels.asSet());
        ConfusionMatrix confusion = new CategoryConfusionMatrix(labels, labelSet);

        populateMatrix(labels.data().toIntArray(), confusion, predictors);
        return confusion;
    }

    public ConfusionMatrix predictMatrix(BooleanColumn labels, NumericColumn... predictors) {
        Preconditions.checkArgument(predictors.length > 0);

        SortedSet<Object> labelSet = new TreeSet<>(labels.asSet());
        ConfusionMatrix confusion = new StandardConfusionMatrix(labelSet);

        populateMatrix(labels.toIntArray(), confusion, predictors);
        return confusion;
    }

    public int[] predict(NumericColumn... predictors) {
        Preconditions.checkArgument(predictors.length > 0);
        int[] predictedLabels = new int[predictors[0].size()];
        for (int row = 0; row < predictors[0].size(); row++) {
            double[] data = new double[predictors.length];
            for (int col = 0; col < predictors.length; col++) {
                data[row] = predictors[col].getFloat(row);
            }
            predictedLabels[row] = classifierModel.predict(data);
        }
        return predictedLabels;
    }

    @Override
    int predictFromModel(double[] data) {
        return classifierModel.predict(data);
    }

    public double logLikelihood() {
        return classifierModel.loglikelihood();
    }

    public double predictFromModel(double[] x, double[] posteriori) {
        return classifierModel.predict(x, posteriori);
    }

    public double predictFromModel(int row, double[] posteriori, NumericColumn... predictors) {
        double[] data = new double[predictors.length];
        for (int col = 0; col < predictors.length; col++) {
            data[row] = predictors[col].getFloat(row);
        }
        return classifierModel.predict(data, posteriori);
    }
}