* @article{Lievens2006,
* author = {S. Lievens and B. De Baets and K. Cao-Van},
* journal = {Annals of Operations Research},
* title = {A Probabilistic Framework for the Design of Instance-Based Supervised Ranking Algorithms in an Ordinal Setting},
* year = {2006}
* }
*
* @phdthesis{Cao-Van2003,
* author = {Kim Cao-Van},
* school = {Ghent University},
* title = {Supervised ranking: from semantics to algorithms},
* year = {2003}
* }
*
* @mastersthesis{Lievens2004,
* author = {Stijn Lievens},
* school = {Ghent University},
* title = {Studie en implementatie van instantie-gebaseerde algoritmen voor gesuperviseerd rangschikken},
* year = {2004}
* }
*
*
*
* Valid options are:
*
* -D * If set, classifier is run in debug mode and * may output additional info to the console* *
-C <REG|WSUM|MAX|MED|RMED> * Sets the classification type to be used. * (Default: MED)* *
-B * Use the balanced version of the Ordinal Stochastic Dominance Learner* *
-W * Use the weighted version of the Ordinal Stochastic Dominance Learner* *
-S <value of interpolation parameter> * Sets the value of the interpolation parameter (not with -W/T/P/L/U) * (default: 0.5).* *
-T * Tune the interpolation parameter (not with -W/S) * (default: off)* *
-L <Lower bound for interpolation parameter> * Lower bound for the interpolation parameter (not with -W/S) * (default: 0)* *
-U <Upper bound for interpolation parameter> * Upper bound for the interpolation parameter (not with -W/S) * (default: 1)* *
-P <Number of parts> * Determines the step size for tuning the interpolation * parameter, nl. (U-L)/P (not with -W/S) * (default: 10)* * * More precisely, this is a simple extension of the OSDLCore class, * so that the OSDLCore class can be used within the WEKA environment. * The problem with OSDLCore is that it implements both *
classifyInstance and distributionForInstance
* in a non trivial way.
*
* One can evaluate a model easily with the method evaluateModel
* from the Evaluation class. However, for nominal classes
* they do the following: they use distributionForInstance
* and then pick the class with maximal probability. This procedure
* is not valid for a ranking algorithm, since this destroys
* the required monotonicity property.
*
* This class reimplements distributionForInstance in the
* following way: first classifyInstance of
* OSDLCore is used and the chosen label then gets
* assigned probability one. This ensures that the classification
* accuracy is calculated correctly, but possibly some other statistics
* are no longer meaningful.
*
classifyInstance from OSDLCore and
* assign probability one to the chosen label.
* The implementation is heavily based on the same method in
* the Classifier class.
*
* @param instance the instance to be classified
* @return an array containing a single '1' on the index
* that classifyInstance returns.
*/
public double[] distributionForInstance(Instance instance) {
// based on the code from the Classifier class
double[] dist = new double[instance.numClasses()];
int classification = 0;
switch (instance.classAttribute().type()) {
case Attribute.NOMINAL:
try {
classification =
(int) Math.round(classifyInstance(instance));
} catch (Exception e) {
System.out.println("There was a problem with classifyIntance");
System.out.println(e.getMessage());
e.printStackTrace();
}
if (Utils.isMissingValue(classification)) {
return dist;
}
dist[classification] = 1.0;
return dist;
case Attribute.NUMERIC:
try {
dist[0] = classifyInstance(instance);
} catch (Exception e) {
System.out.println("There was a problem with classifyIntance");
System.out.println(e.getMessage());
e.printStackTrace();
}
return dist;
default:
return dist;
}
}
/**
* Returns the revision string.
*
* @return the revision
*/
public String getRevision() {
return RevisionUtils.extract("$Revision: 5987 $");
}
/**
* Main method for testing this class and for using it from the
* command line.
*
* @param args array of options for both the classifier
* OSDL and for evaluateModel
*/
public static void main(String[] args) {
runClassifier(new OSDL(), args);
}
}