org.codehaus.jet.regression.estimators
Class OLSMultipleLinearRegressionEstimator

java.lang.Object
  extended by org.codehaus.jet.regression.estimators.AbstractMultipleLinearRegressionEstimator
      extended by org.codehaus.jet.regression.estimators.OLSMultipleLinearRegressionEstimator
All Implemented Interfaces:
MultipleLinearRegressionEstimator

public class OLSMultipleLinearRegressionEstimator
extends AbstractMultipleLinearRegressionEstimator

The OLS implementation of the multiple linear regression OLS assumes the covariance matrix of the error to be diagonal and with equal variance.

 u ~ N(0, sigma^2*I)
 
Estimated by OLS,
 b=(X'X)^-1X'y
 
whose variance is
 Var(b)=MSE*(X'X)^-1, MSE=u'u/(n-k)
 

Author:
Mauro Talevi

Field Summary
 
Fields inherited from class org.codehaus.jet.regression.estimators.AbstractMultipleLinearRegressionEstimator
X, Y
 
Constructor Summary
OLSMultipleLinearRegressionEstimator()
           
 
Method Summary
 void addData(double[] y, double[][] x, double[][] covariance)
          Adds sample and covariance data
protected  org.apache.commons.math.linear.RealMatrix calculateBeta()
          Calculates beta by OLS:
protected  org.apache.commons.math.linear.RealMatrix calculateBetaVariance()
          Calculates the variance on the beta by OLS:
protected  double calculateYVariance()
          Calculates the variance on the Y by OLS:
 
Methods inherited from class org.codehaus.jet.regression.estimators.AbstractMultipleLinearRegressionEstimator
addXSampleData, addYSampleData, calculateResiduals, estimateRegressandVariance, estimateRegressionParameters, estimateRegressionParametersVariance, estimateResiduals
 
Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
 

Constructor Detail

OLSMultipleLinearRegressionEstimator

public OLSMultipleLinearRegressionEstimator()
Method Detail

addData

public void addData(double[] y,
                    double[][] x,
                    double[][] covariance)
Description copied from interface: MultipleLinearRegressionEstimator
Adds sample and covariance data

Parameters:
y - the [n,1] array representing the y sample
x - the [n,k] array representing x sample
covariance - the [n,n] array representing the covariance matrix or null if not appropriate for the specific implementation

calculateBeta

protected org.apache.commons.math.linear.RealMatrix calculateBeta()
Calculates beta by OLS:
 b=(X'X)^-1X'y
 

Specified by:
calculateBeta in class AbstractMultipleLinearRegressionEstimator

calculateBetaVariance

protected org.apache.commons.math.linear.RealMatrix calculateBetaVariance()
Calculates the variance on the beta by OLS:
  Var(b)=(X'X)^-1
 

Specified by:
calculateBetaVariance in class AbstractMultipleLinearRegressionEstimator
Returns:
The beta variance

calculateYVariance

protected double calculateYVariance()
Calculates the variance on the Y by OLS:
  Var(y)=Tr(u'u)/(n-k)
 

Specified by:
calculateYVariance in class AbstractMultipleLinearRegressionEstimator
Returns:
The Y variance


Copyright © 2006-2007. All Rights Reserved.