FYI
public MahalanobisDistanceMeasure(Vector meanVector,Matrix inputMatrix,
boolean inversionNeeded)
{
this.meanVector=meanVector;
if(inversionNeeded)
setCovarianceMatrix(inputMatrix);
else
setInverseCovarianceMatrix(inputMatrix);
}
@Override
public void configure(Configuration job) {
// nothing to do
}
@Override
public Collection<Parameter<?>> getParameters() {
return Collections.emptyList();
}
@Override
public void createParameters(String prefix, Configuration jobConf) {
// nothing to do
}
The wont work in a clustering environment where these class names are
instantiated via reflection. Modify the configure function to read the
deserialized vectors and matrices. See one of the weighted distance measure
classes and see how weight vector is used
Robin
