Implementation of the VFDR algorithm for the Weka machine-learning platform. The algorithm is described in the following paper:

João Gama and Petr Kosina. Learning decision rules from data streams. In Proceedings of the Twenty-Second International Joint Conference on Artificial Intelligence - Volume Two, IJCAI’11, pages 1255–1260. AAAI Press, 2011.

The algorithm performs on-line classification on numeric or nominal attributes. It learns decision rules,can perform multiclass classification, and needs only one pass on the data to build its model. However, this version does not support concept drift.

• Open Weka's package manager (in Weka's GUI chooser, tools > Package manager)
• Scroll to vfdr and click Install

• Assuming weka.jar is in the Java classpath, run

java weka.core.WekaPackageManager -install-package vfdr

• Build the package's zip archive or download it from /releases.
• Assuming weka.jar is in the Java classpath, run

java weka.core.WekaPackageManager -install-package path/to/zip

• Update Description.props
• From the top level directory, execute ant make_package -Dpackage="vfdr-${version}" -buildfile ./build_package.xml, where ${version} is the version number.
• The created zip will live under /dist