how to apply the prior knowledge into data submission of decision tree?

Hello, I am using decision tree for nuclide identification. At the moment, I just simply throw the nuclide spectrum which has 3000 energy bins with instance counts (intneisty) of each bin into decision tree and let computer to generator rules for me. The results are good. However, if two or more nuclide mixed, the spectrums are more complex as some peaks in the spectrum from two or more nuclide could mixed each other. In this case, the deicison tree cannot generate the rules to seperate the mixed nuclides very well. I do have some prior knowledge about the nuclide and mixture of nuclides, i.e. some specific peak for certain nuclide, certain partterns for mixed peaks and etc. Dose knime has any decision tree can process the nuclide spectrum and prior knowledge parallel? Or any plugin can do the job for me? All kind of ideas and experience are welcomed. Thanks.   

One option would be to add a new column upstream of the decision tree learner that expresses this prior knowledge.  Such a column could be generated in a variety of ways including with a rule engine or snippet nodes.  Does that help?





Thanks Aaron. Yes, it is option. however, the prior knowledge is much more complex than add a column as the prior knowledge is energy bin related as well as intensity related. i.e.a simple prior knowledge could be:  bin1, bin50, and bin100 have certain relationship (ratio or sum), then you will be 95% sure there is certain nuclide. So, if a idea or experience about process the those prior knowledge and spectrum parallelly, it might be better. how do you think? Thanks.