Maximum Likelihood Rule Ensembles (MLRules)

Maximum Likelihood Rule Ensembles (MLRules) is a new rule induction algorithm for solving classification problems via probability estimation. The ensemble is built using boosting, by greedily minimizing the negative loglikelihood which results in estimating the class conditional probability distribution. The main advantage of decision rules is their simplicity and comprehensibility: they are logical statements of the form "if condition then decision", which is probably the easiest form of model to interpret. On the other hand, by exploiting a powerful statistical technique to induce the rules, the final ensemble has very high prediction accuracy.

The MLRules algorithm was written in Java and requires that Weka environment is installed on the computer. We provide the following files:

If you want to use MLRules within Weka GUI (or from Weka command-line), you simply need to add mlrules.jar file to the classpath, either by changing the CLASSPATH variable or while invoking Weka from command line, e.g.:

java -cp "weka.jar;mlrules.jar" weka.gui.GUIChooser

Starting with version 3.5.4 one can also add mlrules.jar file permanently in the RunWeka.ini ("cp" option). In case of any problems with adding MLRules, read the following text describing how to add new classifier in the "book version" of Weka.

You can also use MLRules from the command line. Assuming the appropriate jars are in your classpath, by writing:

java weka.classifiers.rules.MLRules

you will obtain a list of options (with their description), both common to Weka environment and specific to our code.

Finally, if you want to use MLRules from the Java code, weka.classifiers.rules.MLRules is the only class that you need to use.

The software is open-source and is provided under GNU General Public Licence (GPL). When using our software for the research purposes, please cite:

@article{DemKotSlo08MLRules,
  author    = {Krzysztof Dembczy\'nski and Wojciech Kot{\l}owski and Roman S{\l}owi\'nski},  
  title     = {Maximum likelihood rule ensembles},
  booktitle = {Proceedings of the 25th International Conference on Machine Learning (ICML 2008)},
  year      = {2008}
}

If you have any problems, do not hesitate to write: .

back