In this recipe, we will see another variant of least squares regression named generalized least squares regression. GLSMultipleLinearRegression
implements Generalized Least Squares to fit the linear model Y=X*b+u.
Create a method that takes a two-dimensional double array, a one-dimensional double array, and a two-dimensional double array for the regression's omega parameter:
public void calculateGlsRegression(double[][] x, double[] y, double[][] omega){
Create a GLS regression object, the data points, and the omega parameter:
GLSMultipleLinearRegression regression = new GLSMultipleLinearRegression(); regression.newSampleData(y, x, omega);
Using the methods of the
GLSMultipleLinearRegression
class, compute various statistics of the regression and finally, close the method:double[] beta = regression.estimateRegressionParameters(); ...