Hi all! I am using Keras library to create a Deep Neural Network for a regression model. I successfully created the workflow, and the model is very accurate and precise. However, I could not find a way to denormalize the predicted column of the output data. Using the node Reference Column splitter I separated the prediction column from the target column, I used Denormalizer node separated for both columns. The node worked well for the target column but did nothing for the predicted one. Does anyone have an explanation or a solution for this problem? Thank you very much.
@Solamento5122 welcome to the KNIME forum. You will temporarily have to rename your predicted column to the original name and rename it back.
3 Likes
That was exactly what I needed to make it work. Now I understand that because the Normalizer node rules were previously applied to the target column before a prediction column was created, when I linked the normalize model output port to the Denormalizer it didn’t find the predicted column in the model and didn’t perform the requested action. Renaming the prediction column temporary as the target one made it possible to apply the rule. To use the denormalized prediction again together with the target column it is necessary to rename the prediction back to its original name. Thank you for helping me to solve this problem and improve my knowledge and understanding of this revolutionary application that is Knime Analytics.
3 Likes
This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.