Classificazione Multi-Classe con classificatori binari + Codice Sorgente

OUTPUT DEL PROGRAMMA E RISULTATI

Un esempio di output del programma è il seguente (Ripetuto per ogni dataset binario):

Esempio Dataset binario 1:

@attribute 5.1 numeric
@attribute 3.5 numeric
@attribute 1.4 numeric
@attribute 0.2 numeric
@attribute Class

@data
4.9,3,1.4,0.2,14.6,3.1,1.5,0.2,15,3.6,1.4,0.2,15.4,3.9,1.7,0.4,15,3.4,1.5,0.2,14.4,2.9,1.4,0.2,14.8,3.4,1.6,0.2,14.8,3,1.4,0.1,14.3,3,1.1,0.1,15.7,4.4,1.5,0.4,15.4,3.9,1.3,0.4,15.1,3.5,1.4,0.3,15.4,3.4,1.7,0.2,15.1,3.7,1.5,0.4,14.6,3.6,1,0.2,15.1,3.3,1.7,0.5,14.8,3.4,1.9,0.2,15,3,1.6,0.2,15,3.4,1.6,0.4,15.2,3.5,1.5,0.2,15.2,3.4,1.4,0.2,14.7,3.2,1.6,0.2,14.8,3.1,1.6,0.2,15.4,3.4,1.5,0.4,15.2,4.1,1.5,0.1,15.5,4.2,1.4,0.2,15,3.2,1.2,0.2,15.5,3.5,1.3,0.2,15.1,3.4,1.5,0.2,15,3.5,1.3,0.3,14.5,2.3,1.3,0.3,15.1,3.8,1.9,0.4,14.8,3,1.4,0.3,14.6,3.2,1.4,0.2,15.3,3.7,1.5,0.2,15,3.3,1.4,0.2,14.7,3.2,1.3,0.2,15.8,4,1.2,0.2,14.4,3.2,1.3,0.2,14.6,3.4,1.4,0.3,15.4,3.7,1.5,0.2,15.7,3.8,1.7,0.3,15.1,3.8,1.5,0.3,14.9,3.1,1.5,0.1,14.9,3.1,1.5,0.1,14.4,3,1.3,0.2,15.1,3.8,1.6,0.2,14.9,3.1,1.5,0.1,15,3.5,1.6,0.6,16.4,3.2,4.5,1.5,06.9,3.1,4.9,1.5,05.5,2.3,4,1.3,06.3,3.3,4.7,1.6,04.9,2.4,3.3,1,06,2.2,4,1,06.1,2.9,4.7,1.4,05.6,2.9,3.6,1.3,06.7,3.1,4.4,1.4,05.6,3,4.5,1.5,05.8,2.7,4.1,1,06.2,2.2,4.5,1.5,05.6,2.5,3.9,1.1,05.9,3.2,4.8,1.8,06.1,2.8,4,1.3,06.3,2.5,4.9,1.5,06.1,2.8,4.7,1.2,06.4,2.9,4.3,1.3,06.6,3,4.4,1.4,06.7,3,5,1.7,05.7,2.6,3.5,1,05.5,2.4,3.7,1,05.8,2.7,3.9,1.2,06,2.7,5.1,1.6,05.4,3,4.5,1.5,06.7,3.1,4.7,1.5,06.3,2.3,4.4,1.3,05.5,2.6,4.4,1.2,06.1,3,4.6,1.4,05.8,2.6,4,1.2,05,2.3,3.3,1,05.6,2.7,4.2,1.3,05.7,3,4.2,1.2,05.7,2.9,4.2,1.3,06.2,2.9,4.3,1.3,06.6,2.9,4.6,1.3,06.8,2.8,4.8,1.4,06,2.9,4.5,1.5,06,3.4,4.5,1.6,05.7,2.8,4.5,1.3,05.2,2.7,3.9,1.4,05,2,3.5,1,05.9,3,4.2,1.5,05.5,2.4,3.8,1.1,05.6,3,4.1,1.3,05.5,2.5,4,1.3,05.1,2.5,3,1.1,05.7,2.8,4.1,1.3,06.5,2.8,4.6,1.5,05.8,2.7,5.1,1.9,07.1,3,5.9,2.1,06.5,3,5.8,2.2,07.6,3,6.6,2.1,04.9,2.5,4.5,1.7,07.3,2.9,6.3,1.8,06.7,2.5,5.8,1.8,06.5,3.2,5.1,2,06.4,2.7,5.3,1.9,06.8,3,5.5,2.1,05.8,2.8,5.1,2.4,06.4,3.2,5.3,2.3,07.7,3.8,6.7,2.2,05.6,2.8,4.9,2,07.7,2.8,6.7,2,06.3,2.7,4.9,1.8,06.7,3.3,5.7,2.1,07.2,3.2,6,1.8,06.1,3,4.9,1.8,06.4,2.8,5.6,2.1,06.4,2.8,5.6,2.2,06.3,2.8,5.1,1.5,06.1,2.6,5.6,1.4,07.7,3,6.1,2.3,06.3,3.4,5.6,2.4,06.4,3.1,5.5,1.8,06,3,4.8,1.8,06.9,3.1,5.4,2.1,06.7,3.1,5.6,2.4,06.9,3.1,5.1,2.3,06.8,3.2,5.9,2.3,06.7,3.3,5.7,2.5,06.5,3,5.2,2,06.2,3.4,5.4,2.3,05.9,3,5.1,1.8,06.5,3,5.5,1.8,06,2.2,5,1.5,06.9,3.2,5.7,2.3,06.2,2.8,4.8,1.8,06.3,2.9,5.6,1.8,07.2,3.6,6.1,2.5,05.7,2.5,5,2,07.2,3,5.8,1.6,07.4,2.8,6.1,1.9,07.9,3.8,6.4,2,05.8,2.7,5.1,1.9,06.7,3,5.2,2.3,06.3,2.5,5,1.9,07.7,2.6,6.9,2.3

Multiclasse:

• @attribute 5.1 numeric
• @attribute 3.5 numeric
• @attribute 1.4 numeric
• @attribute 0.2 numeric
• @attribute Class

{0,1,2}@data4.9,3,1.4,0.2,04.6,3.1,1.5,0.2,05,3.6,1.4,0.2,05.4,3.9,1.7,0.4,05,3.4,1.5,0.2,04.4,2.9,1.4,0.2,04.8,3.4,1.6,0.2,04.8,3,1.4,0.1,04.3,3,1.1,0.1,05.7,4.4,1.5,0.4,05.4,3.9,1.3,0.4,05.1,3.5,1.4,0.3,05.4,3.4,1.7,0.2,05.1,3.7,1.5,0.4,04.6,3.6,1,0.2,05.1,3.3,1.7,0.5,04.8,3.4,1.9,0.2,05,3,1.6,0.2,05,3.4,1.6,0.4,05.2,3.5,1.5,0.2,05.2,3.4,1.4,0.2,04.7,3.2,1.6,0.2,04.8,3.1,1.6,0.2,05.4,3.4,1.5,0.4,05.2,4.1,1.5,0.1,05.5,4.2,1.4,0.2,05,3.2,1.2,0.2,05.5,3.5,1.3,0.2,05.1,3.4,1.5,0.2,05,3.5,1.3,0.3,04.5,2.3,1.3,0.3,05.1,3.8,1.9,0.4,04.8,3,1.4,0.3,04.6,3.2,1.4,0.2,05.3,3.7,1.5,0.2,05,3.3,1.4,0.2,04.7,3.2,1.3,0.2,05.8,4,1.2,0.2,04.4,3.2,1.3,0.2,04.6,3.4,1.4,0.3,05.4,3.7,1.5,0.2,05.7,3.8,1.7,0.3,05.1,3.8,1.5,0.3,04.9,3.1,1.5,0.1,04.9,3.1,1.5,0.1,04.4,3,1.3,0.2,05.1,3.8,1.6,0.2,04.9,3.1,1.5,0.1,05,3.5,1.6,0.6,06.4,3.2,4.5,1.5,16.9,3.1,4.9,1.5,15.5,2.3,4,1.3,16.3,3.3,4.7,1.6,14.9,2.4,3.3,1,16,2.2,4,1,16.1,2.9,4.7,1.4,15.6,2.9,3.6,1.3,16.7,3.1,4.4,1.4,15.6,3,4.5,1.5,15.8,2.7,4.1,1,16.2,2.2,4.5,1.5,15.6,2.5,3.9,1.1,15.9,3.2,4.8,1.8,16.1,2.8,4,1.3,16.3,2.5,4.9,1.5,16.1,2.8,4.7,1.2,16.4,2.9,4.3,1.3,16.6,3,4.4,1.4,16.7,3,5,1.7,15.7,2.6,3.5,1,15.5,2.4,3.7,1,15.8,2.7,3.9,1.2,16,2.7,5.1,1.6,15.4,3,4.5,1.5,16.7,3.1,4.7,1.5,16.3,2.3,4.4,1.3,15.5,2.6,4.4,1.2,16.1,3,4.6,1.4,15.8,2.6,4,1.2,15,2.3,3.3,1,15.6,2.7,4.2,1.3,15.7,3,4.2,1.2,15.7,2.9,4.2,1.3,16.2,2.9,4.3,1.3,16.6,2.9,4.6,1.3,16.8,2.8,4.8,1.4,16,2.9,4.5,1.5,16,3.4,4.5,1.6,15.7,2.8,4.5,1.3,15.2,2.7,3.9,1.4,15,2,3.5,1,15.9,3,4.2,1.5,15.5,2.4,3.8,1.1,15.6,3,4.1,1.3,15.5,2.5,4,1.3,15.1,2.5,3,1.1,15.7,2.8,4.1,1.3,16.5,2.8,4.6,1.5,15.8,2.7,5.1,1.9,27.1,3,5.9,2.1,26.5,3,5.8,2.2,27.6,3,6.6,2.1,24.9,2.5,4.5,1.7,27.3,2.9,6.3,1.8,26.7,2.5,5.8,1.8,26.5,3.2,5.1,2,26.4,2.7,5.3,1.9,26.8,3,5.5,2.1,25.8,2.8,5.1,2.4,26.4,3.2,5.3,2.3,27.7,3.8,6.7,2.2,25.6,2.8,4.9,2,27.7,2.8,6.7,2,26.3,2.7,4.9,1.8,26.7,3.3,5.7,2.1,27.2,3.2,6,1.8,26.1,3,4.9,1.8,26.4,2.8,5.6,2.1,26.4,2.8,5.6,2.2,26.3,2.8,5.1,1.5,26.1,2.6,5.6,1.4,27.7,3,6.1,2.3,26.3,3.4,5.6,2.4,26.4,3.1,5.5,1.8,26,3,4.8,1.8,26.9,3.1,5.4,2.1,26.7,3.1,5.6,2.4,26.9,3.1,5.1,2.3,26.8,3.2,5.9,2.3,26.7,3.3,5.7,2.5,26.5,3,5.2,2,26.2,3.4,5.4,2.3,25.9,3,5.1,1.8,26.5,3,5.5,1.8,26,2.2,5,1.5,26.9,3.2,5.7,2.3,26.2,2.8,4.8,1.8,26.3,2.9,5.6,1.8,27.2,3.6,6.1,2.5,25.7,2.5,5,2,27.2,3,5.8,1.6,27.4,2.8,6.1,1.9,27.9,3.8,6.4,2,25.8,2.7,5.1,1.9,26.7,3,5.2,2.3,26.3,2.5,5,1.9,27.7,2.6,6.9,2.3,2La

La precisione (precisione) e il recall (sensibilità) sono due indicatori usati nel machine learning per valutare la qualità di un modello decisionale o di un modello predittivo.

Gli indici di maggior rilevanza sono la Precision e la Recall:

• La precisione è il rapporto tra il numero delle previsioni corrette di un evento (classe) sul totale delle volte che il modello lo prevede. PPV = TP/(TP + FP)
• Il richiamo (recall) misura la sensibilità del modello. È il rapporto tra le previsioni corrette per una classe sul totale dei casi in cui si verifica effettivamente. R = TP/(TP + FN)

La ROC area/curva ROC mette a confronto il TP rate (asse delle ordinate) e FP rate (asse delle ascisse); il calcolo della sua sotto area è utile per valutare l'abilità del classificatore nel separare le istanze positive da quelle negative. Nel caso in questione siamo vicini al valore 1 (il test risulta accurato).

CODICE JAVA

Di seguito il codice java delle...

<classi>
    <WekaUtils>
        package weka.api;
        
        import weka.attributeSelection.BestFirst;
        import weka.attributeSelection.CfsSubsetEval;
        import weka.attributeSelection.InfoGainAttributeEval;
        import weka.attributeSelection.Ranker;
        import weka.classifiers.Evaluation;
        import weka.classifiers.meta.AttributeSelectedClassifier;
        import weka.classifiers.trees.J48;
        import weka.core.Instances;
        import weka.core.converters.ArffSaver;
        import weka.core.converters.CSVLoader;
        import weka.filters.Filter;
        import weka.filters.supervised.instance.SMOTE;
        import weka.filters.unsupervised.attribute.Add;
        import weka.filters.unsupervised.attribute.Remove;
        import java.io.File;
        
        public class WekaUtils {
        
            public static Instances[] elaboraNuoviDataset() throws Exception {
                Instances[] listOfInstances0 = new Instances[4];
                Instances[] listOfInstances1 = new Instances[4];
                Instances[] listOfInstances2 = new Instances[4];
                
                for(int i=1; i<=3; i++) {
                    CSVLoader loader = new CSVLoader();
                    String inputPath = 

"C://Users//MYPC//Desktop//ClassificazioneBinaria//input//" + i + ".csv";
loader.setSource(new File(inputPath));
Instances data = loader.getDataSet();
if (data.classIndex() == -1)
    data.setClassIndex(data.numAttributes() - 1);
Instances filteredData = applicaFiltroRimuovi(data);
filteredData = aggiungiAttributo(filteredData, "last", "0,1", "Class");
filteredData = setValues(filteredData, 0);
filteredData.setClassIndex(filteredData.numAttributes() - 1);
listOfInstances0[i] = new Instances(filteredData);
filteredData = setValues(filteredData, 1);
filteredData.setClassIndex(filteredData.numAttributes() - 1);
listOfInstances1[i] = new Instances(filteredData);
filteredData = applicaFiltroRimuovi(filteredData);
filteredData = aggiungiAttributo(filteredData, "last", "0,1,2", "Class");
for (int j = 0; j < filteredData.numInstances(); j++) {
    filteredData.instance(j).setValue(filteredData.numAttributes() - 1,

i-1);
}
filteredData.setClassIndex(filteredData.numAttributes() - 1);
listOfInstances2[i] = new Instances(filteredData);
}

Instances[] newListOfInstances = new Instances[4];
newListOfInstances[0] = new Instances(listOfInstances2[1]);
newListOfInstances[1] = new Instances(listOfInstances1[1]);
newListOfInstances[2] = new Instances(listOfInstances1[2]);
newListOfInstances[3] = new Instances(listOfInstances1[3]);

newListOfInstances[0].addAll(listOfInstances2[2]);
newListOfInstances[0].addAll(listOfInstances2[3]);
newListOfInstances[1].addAll(listOfInstances0[2]);
newListOfInstances[1].addAll(listOfInstances0[3]);
newListOfInstances[2].addAll(listOfInstances0[1]);
newListOfInstances[2].addAll(listOfInstances0[3]);
newListOfInstances[3].addAll(listOfInstances0[1]);
newListOfInstances[3].addAll(listOfInstances0[2]);

salvaInFileArff(newListOfInstances);
return newListOfInstances;
}

public static void salvaInFileArff(Instances[] listOfIstances) throws Exception {
for(int i=0; i<listOfIstances.length; i++)

{String pathName = i == 0 ? "Multiclasse" : "C"+i;String outputPath ="C://Users//MYPC//Desktop//ClassificazioneBinaria//output//&