Question

我用来训练模型并使用模型再次分类。

我正确地得到了第一部分的统计数据，但没有得到第二部分的统计数据。它再次评估时会给出nullPointerException。我尝试过所有类型的操作，比如在代码等中创建的一个实例上测试它。

java.lang.NullPointerException
        at weka.classifiers.trees.m5.M5Base.classifyInstance(M5Base.java:514)
        at wekaTest.<init>(wekaTest.java:44)
        at wekaTest.main(wekaTest.java:71)

我写的代码片段是：

wekaTest()
{
    try
    {
        FileReader reader = new FileReader("3.arff"); 
        Instances instances = new Instances(reader); 

        // Make the last attribute be the class 
        int numAttr = instances.numAttributes();
        instances.setClassIndex( numAttr - 1);
        M5P tree = new M5P();
        Evaluation eval = new Evaluation(instances);
        eval.crossValidateModel(tree, instances, 10, new Random(1));
        System.out.println(eval.toSummaryString("\nResults\n======\n", false));
        weka.core.SerializationHelper.write("/path/tree.model", tree);
        reader.close();

        FileReader reader2 = new FileReader("3.arff"); 
        Instances instances2 = new Instances(reader2);
        instances2.setClassIndex(instances2.numAttributes() - 1);
        reader2.close();

        Instances labeled = new Instances(instances2);
        Classifier cls = (Classifier) weka.core.SerializationHelper.read("/path/tree.model");

        //instances2.deleteAttributeAt(numAttr-1);
        for(int j=0; j<instances2.numInstances() ;j++)
        {
                //instance temp = new instance(instances2.instance(j));
                //instances2.instance(j).setValue(numAttr-1,-1);
                System.out.println("The instance: " + instances2.instance(j)); 
                double clsLabel = tree.classifyInstance(instances2.instance(j));
                labeled.instance(j).setClassValue(clsLabel);    
        }
    } 
    catch(Exception ex) { ex.printStackTrace(); }
}

Answer 1

可能是您正在编写的树尚未初始化。

Answer 2

谢谢Aditya。其实你是对的！当我在10次交叉验证后编写它时，该变量尚未初始化。

摘录如下：

  try
    {
        FileReader reader2 = new FileReader("3.arff"); 
        Instances instances2 = new Instances(reader2);
        instances2.setClassIndex(instances2.numAttributes() - 1);
        reader2.close();
        int numAttr = instances2.numAttributes();

        Instances labeled = new Instances(instances2);
        Classifier cls = (Classifier) weka.core.SerializationHelper.read("/home/sumit/Desktop/weka test/tree.model");
        cls.setDebug(true);

        Instance inst = new Instance(4);
        inst.setValue(0, instances2.instance(0).value(0));
        inst.setValue(1, instances2.instance(0).value(1));
        inst.setValue(2, instances2.instance(0).value(2));
        inst.setValue(3, -1);
        double clsLabelTest = cls.classifyInstance(inst);
        System.out.println(clsLabelTest);

        //instances2.deleteAttributeAt(numAttr-1);
        for(int j=0; j<instances2.numInstances() ;j++)
        {
                //instance temp = new instance(instances2.instance(j));
                instances2.instance(j).setValue(numAttr-1,-1);
                //System.out.println("The instance: " + instances2.instance(j)); 
                double clsLabel = cls.classifyInstance(instances2.instance(j));
                labeled.instance(j).setClassValue(clsLabel);
        }
        BufferedWriter writer = new BufferedWriter(new FileWriter("/home/sumit/Desktop/weka test/labeled.arff"));           
        writer.write(labeled.toString());
        writer.newLine();
        writer.flush();
        writer.close();
        // Test the model
        //Evaluation eTest = new Evaluation(instances2);
        //eTest.evaluateModel(cls, instances2);
    }

分类时Weka nullPointerException

2 个答案: