Good evening, I just want to clarify the logic of learner:

File: MyFilteredLearner.java

In main method, you said that "evaluation must be done before training".
I ask as I see that the evaluate() and learn() method have partial same code as shown in this snippet:

trainData.setClassIndex(0);
filter = new StringToWordVector();
filter.setAttributeIndices("last");
classifier = new FilteredClassifier();
classifier.setFilter(filter);
classifier.setClassifier(new NaiveBayes());

It looks like when evaluating the classifier, we haven't built the classifier yet.
So, what classifiers are to be evaluated?

Thx.

I'm using Weka 3.6 and while I'm trying to use this code it gives me the error

DenseInstance instance = new DenseInstance(2);
                ^
symbol:   class DenseInstance
location: class Classifier
Classifier.java:39: error: cannot find symbol
DenseInstance instance = new DenseInstance(2);

In-fact DenseInstance class is missing even from the documentation that shipped with my version of Weka.