I am currently performing experiments on text
classification using Weka,
and I am ionterested on using Stacked Generalization applied to an
heuristic classifier I have bilt and some other classification schemas. I
have read in the papers suggested by the Weka book and documentation
(including the paper by Ting and Witten on "SG: when does it work?") that
the level-1 generalizer would optimally be a linear model. Among the Weka
classifiers able to induce linear models, which ones would you suggest as
level-1 learners for Stacking?
If you require accurate class probability estimates, you might want
to use Logistic.java (i.e. logistic regression). If you are interested
in accuracy, you might want to try SMO.java (i.e. a linear support
vector machine). However, it would probably make sense to perform
experiments with those two as well as LinearRegression.java (using
ClassificationViaRegression.java) and LogitBoost.java.