[kepler-users] Weka actors in Kepler

Peter Reutemann fracpete at waikato.ac.nz
Mon Sep 15 02:32:21 PDT 2008


Hi there

I've created a custom IzPack installer of Kepler that includes actors
for the machine learning workbench Weka (Weka homepage
http://www.cs.waikato.ac.nz/ml/weka/). This installer is based on
Ptolemy II rev. 50605, Kepler rev. 7889 and Weka CVS HEAD as of
12/09/2008 (basically a slightly bug-fixed 3.5.8 release).

For those who aren't familiar with Weka, here's a short abstract from
the Weka homepage (for more information, see the Weka homepage,
section "Documentation"):
"Weka is a collection of machine learning algorithms for data mining
tasks. The algorithms can either be applied directly to a dataset or
called from your own Java code. Weka contains tools for data
pre-processing, classification, regression, clustering, association
rules, and visualization. It is also well-suited for developing new
machine learning schemes. "

Currently, the following functionality is available in this custom Kepler build:
- associators (Apriori, Tertius, ...)
- classifiers (decision trees, support vector machines, linear regression, ...)
- clusterers (EM, K-Means, ...)
- attribute selection schemes (attribute ranking, subset evaluation)
- artificial data generators
- preprocessing via various filters (supervised, unsupervised)
- reading from/writing to data files (various formats: ARFF, CSV, ...)
- reading from/writing to databases
- serializing/deserializing of models (classifiers, clusterers)
- evaluation:
  ~ explicit test set
  ~ random train/test split
  ~ cross-validation
  ~ experiments (using the simple setup of the Weka Experimenter)
- visualization
  ~ ROC curves
  ~ cost curves
  ~ tree visualizer (e.g., decision trees)
  ~ graph visualizer (e.g., bayesian networks)
  ~ attribute summary

You can download the custom installer (or just the actors) from the
following URL ("Kepler" section):
  http://www.cs.waikato.ac.nz/~fracpete/downloads/

A few example workflows and screenshots are available from this
location as well.

Cheers, Peter
-- 
Peter Reutemann, Dept. of Computer Science, University of Waikato, NZ
http://www.cs.waikato.ac.nz/~fracpete/ Ph. +64 (7) 858-5174



More information about the Kepler-users mailing list