Method get2ClassLinear2D

JSAT/test/jsat/FixedProblems.java:57–70

Creates a 2D linearly separable problem @param dataSetSize0 size of the first class @param dataSetSize1 size of the second class @param sep the separation between the two classes. The true decision boundary stays in the same location regardless of this value @param rand source of randomness @return

(int dataSetSize0, int dataSetSize1, double sep, Random rand)

     * @return a 2d testing set
     */
    public static ClassificationDataSet get2ClassLinear2D(int dataSetSize0, int dataSetSize1, double sep, Random rand)
    {
        ClassificationDataSet train = new ClassificationDataSet(2, new CategoricalData[0], new CategoricalData(2));
        
        NormalM a = new NormalM(DenseVector.toDenseVec(sep, sep), Matrix.eye(2));
        NormalM b = new NormalM(DenseVector.toDenseVec(-sep, -sep), Matrix.eye(2));
        
        for(Vec s : a.sample(dataSetSize0, rand))
            train.addDataPoint(s, new int[0], 0);
        for(Vec s : b.sample(dataSetSize1, rand))
            train.addDataPoint(s, new int[0], 1);
        
        return train;
    }
    
    /**
     * Generates a linearly separable binary classification problem