I'm working on single-label multi-class classification task using weka API. I want to train one Multilayer Perceptron model to predict the class percentage values for each class by generating one-hot encoded columns from my class attribute column.
In the examples on the Internet, after generating an Instances object in the Weka Java API, only one attribute is selected as a class attribute.
On the examples:
Data upload
Instances data = ...
Selecting the last attribute as a class attribute
data.setClassIndex (data.numAttributes () - 1);
What I'm looking for:
Selecting multiple attributes as a class attribute
Instances data = ...
data.setClassIndex (1,2,5,6,...);
How can I do that?
