Based on my research it seems that there is no way to automatically export a naive bayes model to pmml. I would like to export my multinomial naive bayes model into this pmml format and so I have resorted to trying to parse it myself. My problem is that I don't completely understand how I can output the values in the format required by the PMML format. I am training a text classification model which has binary features 0 or 1 as to whether that word is in the text. I then train the model using multinomial naive bayes.
Based on my understanding of the PMML format, for each feature x_i you are supposed to list the number of times each value of x_i occured in each class C_j so for one of my features I would have a count of the number of times x_i=0 occured for each class and the number of times x_i=1 occured for each class. Based on my understanding of the multinomial Naive Bayes model it seems that you can only get the actually probabilities p(x_i|C_j) and not counts. My questions are 1) Is that correct? If so are there ways of getting such counts? I am confused about how one can actually convert a Naive Bayes model into the PMML format if that information does not get persisted to the model which confuses me about the Py2PMML convertor. Chirag ------------------------------------------------------------------------------ Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server from Actuate! Instantly Supercharge Your Business Reports and Dashboards with Interactivity, Sharing, Native Excel Exports, App Integration & more Get technology previously reserved for billion-dollar corporations, FREE http://pubads.g.doubleclick.net/gampad/clk?id=157005751&iu=/4140/ostg.clktrk _______________________________________________ Scikit-learn-general mailing list Scikit-learn-general@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
