Weka Guide

Weka is installed in /usr/local/weka in our department machines.
For those who want to install Weka on your machines, you can simply unzip the weka after downloading it from http://www.cs.waikato.ac.nz/ml/weka/.

Weka provides the Explorer and Experimenter GUI that can be run with command:

    java -Xmx1024M -jar weka.jar

The -Xmx1024M option for maximum heap size ensures your run to have enough memory.
You can set this option in the following command-line as well for handling large data.

For in-depth usage, Weka provides the command line interface, which is recommended because of memory and functionality.
After setting the CLASSPATH to /usr/local/weka/weka.jar or passing -cp option, you can train C4.5 decision tree as follows:

    java weka.classifiers.trees.J48 -t /usr/local/weka/data/weather.numeric.arff

For experiments, you can write a bash/csh script with multiple runs including similar to the previous lines.

Weka package contains the documents in it: WekaManual.pdf and documentation.html.
For the package reference, you can take a look at the Package Documentation from documentation.html or /usr/local/weka-3-7-10/doc/index.html.

Here is the quick introduction to Weka.