Decision trees are popular tools for classification. Here you can download a program that can build univariate, binary decision trees. The code is written in Java (platform 5.0). The code can be freely used for research purposes.
The zip file contains the package bodon and its subpackages. The package bodon.mining.classification.decisiontree contains the bytecodes of the decision tree building program. This package use many useful methods that are not part of the standard Java library, hence they are placed in the package bodon.common. Since the bytecodes are not in one (the default) package it is necessary to specify the classpath. To run the program do the following:
unzip bodon_package.zip java -classpath . bodon.mining.classification.decisiontree.BuildDTModel
The package also contains a program that is able to predict a classvalue of an unknown point. To run this program type the following command.
java -classpath . bodon.mining.classification.decisiontree.PredictUnseenPoint
To compile do the following:
unzip bodon_package_src.zip make
You can also generate a local copy of the documentation. For this download bodon_package_src.zip and then:
unzip bodon_package_src.zip make doc
You will find the documentation in directory doc.
The project was started in Eclipse. Since I don't like mouse clicks (they are too slow) I changed the development environment to Emacs using the JDEE package. Emacs with jdee is an excellent Java development environment, I propose it to everybody.