使用Java中的随机森林打印实际和预测的类标签

Question

我有一个带有10000条记录的大数据集，因此5000属于1类，剩余5000属于-1类。我使用了随机森林，并获得了超过90％的良好精度。

现在，如果我有一个arff文件

@relation cds_orf
@attribute start numeric
@attribute end numeric
@attribute score numeric
@attribute orf_coverage numeric
@attribute class {1,-1}
@data
(suppose this contains 5 records)

我的输出应该像这样

 No   Actual_class   Predicted class
 1     1                   1
 2     1                   1   
 3    -1                  -1  
 4     1                   -1
 5     1                    1

我希望Java代码打印此输出。谢谢。(注意：我使用了classifier.classifyInstance((，但它给出了nullpointerexception(

Answer 1

好吧，我在大量研究后发现了答案。以下代码执行相同的操作，并将输出写入磁盘文件ORF_OUT。

import java.io.BufferedReader;
import java.io.BufferedWriter;
import java.io.FileReader;
import java.io.FileWriter;
import java.io.PrintWriter;
import java.util.Random;
import weka.classifiers.Evaluation;
import weka.classifiers.trees.RandomForest;  
import weka.core.Instances;
 
/**
 *
 * @author samy
 */
public class WekaTest {
 
    /**
     * @throws java.lang.Exception
     */
    public static void rfnew() throws Exception {
        BufferedReader br;
        int numFolds = 10;
        br = new BufferedReader(new FileReader("orf_arff"));
 
        Instances trainData = new Instances(br);
        trainData.setClassIndex(trainData.numAttributes() - 1);
        br.close();
        
        RandomForest rf = new RandomForest();
        rf.setNumTrees(100);         
     
        Evaluation evaluation = new Evaluation(trainData);
        evaluation.crossValidateModel(rf, trainData, numFolds, new Random(1));
        rf.buildClassifier(trainData);
        PrintWriter out = new PrintWriter("orf_out");
        out.println("No.tTruetPredicted");
        for (int i = 0; i < trainData.numInstances(); i++)      
        {
            String trueClassLabel;
            trueClassLabel = trainData.instance(i).toString(trainData.classIndex());
             // Discreet prediction
            double predictionIndex = 
            rf.classifyInstance(trainData.instance(i)); 
            // Get the predicted class label from the predictionIndex.
            String predictedClassLabel;            
            predictedClassLabel = trainData.classAttribute().value((int) predictionIndex);
            out.println((i+1)+"t"+trueClassLabel+"t"+predictedClassLabel);
        }
        
        out.println(evaluation.toSummaryString("nResultsn======n", true));
        out.println(evaluation.toClassDetailsString());
        out.println("Results For Class -1- ");
        out.println("Precision=  " + evaluation.precision(0));
        out.println("Recall=  " + evaluation.recall(0));
        out.println("F-measure=  " + evaluation.fMeasure(0));
        out.println("Results For Class -2- ");
        out.println("Precision=  " + evaluation.precision(1));
        out.println("Recall=  " + evaluation.recall(1));
        out.println("F-measure=  " + evaluation.fMeasure(1)); 
        out.close();
    }
}

我需要在代码中使用buildClassifier。