classcmar_app.java

来自「多关联分类算法」· Java 代码 · 共 86 行

/* -------------------------------------------------------------------------- */
/*                                                                            */
/*                APRIORI-TFP CMAR (CLASSIFICATION BASED ON                   */
/*                 MULTIPLE ASSOCIATION RULES) APPLICATION                    */
/*                                                                            */
/*                             Frans Coenen                                   */
/*                                                                            */
/*                           Friday 5 March 2004                              */
/*                                                                            */
/*                      Department of Computer Science                        */
/*                        The University of Liverpool                         */
/*                                                                            */
/* -------------------------------------------------------------------------- */

import java.io.*;

/* Classification application the CMAR (Classification based on Multiple 
Associate Rules) algorithm proposed by Wenmin Li, Jiawei Han and Jian Pei,
but founded on Apriori-TFP.

Compile using:

javac ClassCMAR_App.java

Run using the java interpreter, Example:

java ClassCMAR_App -FpimaIndians.D42.N768.C2.num -N2 -S1 -C50

(-F filename, -N number of classifiers).              */

public class ClassCMAR_App {

    // ------------------- FIELDS ------------------------

    // None

    // ---------------- CONSTRUCTORS ---------------------

    // None

    // ------------------ METHODS ------------------------

    public static void main(String[] args) throws IOException {
	double time1 = (double) System.currentTimeMillis();
	
	// Create instance of class ClassificationPRM	
	AprioriTFP_CMAR newClassification = new AprioriTFP_CMAR(args);
				
	// Read data to be mined from file (method in AssocRuleMining class)
	newClassification.inputDataSet();
	
	// Reorder input data according to frequency of single attributes
	// excluding classifiers. Proceed as follows: (1) create a conversion
	// array (with classifiers left at end), (2) reorder the attributes 
	// according to this array. Do not throw away unsupported attributes 
	// as when data set is split (if distribution is not exactly even) we 
	// may have thrown away supported attributes that contribute to the 
	// generation of CRs. NB Never throw away classifiers even if
	// unsupported!
	newClassification.idInputDataOrdering();  // ClassificationAprioriT
	newClassification.recastInputData();      // AssocRuleMining
	
	// Create training data set (method in ClassificationAprioriT class)
	// assuming a 50:50 split
        newClassification.createTrainingAndTestDataSets();
	
	// Mine data, produce T-tree and generate CRs
	double accuracy = newClassification.startCMARclassification();
	newClassification.outputDuration(time1,
				(double) System.currentTimeMillis());
	
	// Output
	//newClassification.outputFrequentSets();
	newClassification.outputNumFreqSets();
	newClassification.outputNumUpdates();
	newClassification.outputStorage();
	//newClassification.outputTtree();
	System.out.println("Accuracy = " + accuracy);
	newClassification.getCurrentRuleListObject().outputNumCMARrules();
	newClassification.getCurrentRuleListObject().outputCMARrules();
	
	// End
	System.exit(0);
	}
    }