How to add weka features in a new algorithm?

问题:

I want to add a new algorithm to weka with features of classification, clustering, association etc in one algo. How should I write a code to include all the weka features and add a tab to weka for this new algorithm. I have added a dummy algorithm to weka and it works now I want to add an algorithm which has combination of features of weka.

Thanks

回答1:

If you want to add a new algorithm in Weka, have a look at the Weka Manual ( http://www.cs.waikato.ac.nz/ml/weka/index.html )

In the part IV - Appendix, you have the chapter Extending Weka and inside the part Writing a new Classifier. Very basically, you have to extend a Classifier like AbstractClassifier or RandomizableClassifier.

In the case of clustering algorithms look at the part Other Algorithms. You will have to extend, for instance, the class weka.clusterers.AbstractClusterer.

But, if you want to use existing algorithms in your own program, have a look to this link: http://weka.wikispaces.com/How+do+I+use+WEKA%27s+classes+in+my+own+code%3F