I'm quite new with KNIME. My questions is: I ran a decision tree with a credit dataset. I've built a project with the classical nodes (file reader, partitioning , decisision tree, scorer) and executed. The decision variable is a nominal one with the values "good" or "bad" according to the credit check results.
How can run the same classification tecnique on a new dataset where I do not have the "good" / "bad" variable set. In other words how can I apply the model to new data?
Have you tried running the predictor without it and got an error?
You don't need to target column when predicting, just connect the Decision Tree Predictor node to the output of the Decision Tree Learner and the data input only needs to have the descriptor columns, you don't need the target column.
For example, if I had the following dataset
RowID: var1, var2, var3, var4, credit score (target class)
I use these columns for training the decision tree, I only need the following for predicting:
RowID: var1, var2, var3, var4
The credit score (what you are predicting) does not need to be included to make a prediction as it is not used to as a split point in the tree.