about seperation of modeling and prediction

lushawang

I used cross validation operator for data mining, including SVM operator and apply model operator. It always finishes in a couple of minutes. Now I have to separate the modeling part from the prediction one. So I simply use SVM classifier and then write model to a file in one process. While in the other process, I read the model from the output file and then apply the model for the test data. However, the second process runs really slow. Besides, a bunch of warning showed up, like some specific attribute is not contained in the example set.

I really appreciate any of your help.

meliniak

is there any preprocessing done in the modelling process? like attribute changing so it can go into the svm operator? if so, same operators have to be applied before the model is.