Continuous prediction models with numerical values

I’ve made several attempts to create continuous prediction models using the SVM learner, as well as the neural network nodes. Based on the output that I have seen, it does not appear that I am getting interpolation of results, only the result closest to the training set. Am I doing something wrong, or am I asking too much from these nodes?


I’m not sure I fully understand the question but I beleive that only the RProp MLP Learner handles a numberic target.

All the best,


the RProp MLP learner and the MLP Predictor in KNIME can handle numerical target attributes. Just make sure that they are normalized in the range [0,1].
The KNIME SVM does not support regression. The fact that you are getting results at all suggests that you have transformed the DataType from your target column from double to String/nominal?
If you want to use a SVM for regression, you can install the Weka-integration and have a look at the SMOreg classifier (in Classification Algorithms->Functions).

Best regards