work with pnn

Hello again, I’m looking for some examples on how to use RadialBasisFunctionPredictorRow. I have to use this class on my java source code, and I really don’t know how to start.

Regards again.

Pablo Fortunato

Hello, interesting that someone uses this class :wink: However, the best way is to start from the corresponding NodeModel called RadialBasisFunctionPredictorNodeModel. If you have a look into the execute method, you will find out that this predictor rows are created by a BasisFunctionPredictorCellFactory (inside the abstract class BasisFunctionPredictorNodeModel). That’s all. If you tell me a little bit more what are you intending to implement, I would be happy to support you with additional details.
Best, Thomas

Hi Thomas. What I have to do is the next. From an .arff file, create a pnn model. Then use that model to predict some data. If you know about weka, well, I used it for the same thing, but I’ve found that knime is more efficient and take less time to build an model.
I will follow your advices about the RadialBasisFunctionPredictorNodeModel.

Thanks again.


Hi Pablo, are you trying to adapt the RadialBasisFunctionPredictor source code in order to fit your needs? It sounds like if you have already created a flow with ARFF-Reader, RadialBF-Learner and -Predictor. I am just curious, what else is missing to complete the flow…
Cheers, Thomas

Hi, in first place, thanks for your answer.
I don’t want to change the source code of knime class (I guess :slight_smile: ). I want an instance of RadialBasisFunctionPredictor in my source code. With weka models, I can create a (for example) MultilayerPerceptron model and execute the classifyInstance method to obtain the data that I need.
With Knime, is that possible? and if it’s true (or yes), How can I do that?
Also I saw in forum, that I could use the workflow file created. Inside my java code, How can I execute that workflow. I’m trying with executeAll (of course) but how can I see the results of that execution?
Thanks again, again.