Polynomial Regression Learner

How can I use a learned Polynomial Regression model on a training data set for prediction of a test data set ?

I tried instantiating two Polynomial Regression Learners in series, the first one connected to the training data set, and with its Regression Coefficients output (Blue square) connected to the PMML preprocessing input for the second Polynomial Regression learner, whose input "training data" port is connected to the test data set. In other words, I want the first Polynomial Regression Learner to be the learner, and the second one to be the predictor. However, this arrangement causes the second Polynomial Regression learner to also try to learn from the test data, which I do not want. I want to merely use the trained polynomial regression model purely as a predictor of the test data (i.e. by just plugging in the test data values and measuring error). How do I achieve this?




Use the Regression Predictor node.

Thank you very much. It works for me.

Two followup questions:

For time series prediction, does Knime provide any learner/predictor models besides Polynomial/Linear Autoregressive and RPROP (backprop) neural networks (with MLP Predictor) ? If so, I would like to use them in an ensemble.

Also, are there any pre-built ensemble models for time series prediction in Knime ? I don't mind building them manually by combining tables of predicted data from multiple models, but knowing whether something like this already exists or not would be helpful.