Decision Tree/Neural Network for two columns


I am just starting with the KNIME. I have a question which i cannot solve it. I wanna build a model using decision tree or neural network for 2 columns and than see the correlation between them. Is it possible and how i can do it?


See attached for a simple workflow demonstrating some options here. Hopefully it provides some inspiration.





Hi Aaron!


Thanks for the workspace. I have one more question, i cannot execute my MultiLayerPerceptron Predictor. I get the message:

ERROR     MultiLayerPerceptron Predictor     Execute failed: cannot be cast to

What can be the reason of that?





Hi Malvina, 

This error message informs you about the right datatype for the target column and input columns in the MLP algorithm 

You have to change your target column datatype to string to use MLP model (with "string to number" node). 

Also you can try with other algorithms that allow you to work with numerical data (PNN model may be another neural network implementation)





Thanks for the advice. I have found the mistake. I forgot to use "string to number" node between "Partitioning" and "MultiLayerPerceptron Predictor" (I have only used it bewteen "Partitioning" and "RProp MLP Learner"), so the data didnt match.


I do have one more problem. The accuracy of my model is very low (around 20%), and with increasing number of loops it decreases!

I have no idea whats wrong with it. Is there a problem with Loop node or with the model that i have choosed?