Re: What is the difference between the M5Rules and M5P algorithms ?
M5Rules generates a series of M5 trees, where only the "best" (highest coverage) leaf/rule is retained from each tree. At each stage, the instances covered by the best rule are removed from the training data before generating the next tree. The algorithm is similar to the PART method for classification trees, except that always builds a full tree at each stage and does not employ the partial tree building speed-up of PART. M5P builds a single decision tree. It is certainly possible that an M5 rules classifier could outperform M5P on a given dataset.
Is it possible to know what is the difference between the M5Rules and the
M5P algorithms in Weka? Cause I read the referenced papers on the subject
(papers below) and I still don't understand exactly the difference. In my
understanding: the M5P algorithm must be better than the M5Rules, but in my
results: the M5Rules is better, so....
Thank you for your help !
Geoffrey Holmes, Mark Hall, Eibe Frank: Generating Rule Sets from Model
Trees. In: Twelfth Australian Joint Conference on Artificial Intelligence,
Ross J. Quinlan: Learning with Continuous Classes. In: 5th Australian Joint
Conference on Artificial Intelligence, Singapore, 343-348, 1992.
Y. Wang, I. H. Witten: Induction of model trees for predicting continuous
classes. In: Poster papers of the 9th European Conference on Machine