I'm trying to compare the goodness of different classifiers. I'm using Knime and several mining algorithms, like Decision Tree, Naive Bayes, Fuzzy Rules, SVN, and Probabilistic Neural Network. For all these algorithms, Knime provides both a learner as well as a predictor. I also need to do the same with the Association Rules learner, as I'm interesting in the knowledge to be extracted through them. I have prepared all input data as binary vectors. However, it seems that Knime does not provide a predictor for the Association Rules it extracts. Is that correct? Why does it provide it for all other algorithms I mentioned, but not for Association Rules? Many thanks Jesús
Correct, all mining methods are separated into learner and predictor making it more flexible when applying new unknown data to an already generated model. In case of the association rule learner, the Subset Matcher node is the one to be used on association rules together with frequent item sets.