OneRule.jl

Implementation of the 1-Rule data mining algorithm by Robert Holte ("Very simple classification rules perform well on most commonly used datasets" in: Machine Learning 11.1 (1993), pp. 63-90) using the Julia programming language.

For more information about the algorithm see e.g.: Machine Learning - (One|Simple) Rule, OneRClassifier - One Rule for Classification or have a look at

Witten, Ian H., Eibe Frank, and Mark A. Hall. Data Mining Practical Machine Learning Tools and Techniques Third Edition. Morgan Kaufmann, 2017, pp. 93-96

The implementation of this algorithm has started as an example for a university course in data science at Baden-Württemberg Cooperative State University Karlsruhe.

It has beend extended to offer full functionality (e.g. for predicting) and it has been adapted to conform to common ML interfaces. It's now part of the meta-package MLJ.jl).

The model works on categorical data for the features as well as the target class for training (using the get_best_tree function.) You find an example on how to use it in the examples directory.

For a description on its use within the context of the MLJ package, just type

using MLJ
doc("OneRuleClassifier", pkg="OneRule")

The core algorithm used to explain the inner workings of the OneRule model in the above-mentioned course, can still be found in the branch teaching.