Classification Based on Association Rules

Provides the infrastructure for association rule-based classification including the algorithms CBA, CMAR, CPAR, C4.5, FOIL, PART, PRM, RCAR, and RIPPER to build associative classifiers.

This R package is an extension of the package arules to perform association rule-based classification. It includes currently two classification algorithms. The first is the CBA algorithm described in Liu, et al. 1998. The second is a new weighted majority-vote based algorithm called bCBA which is currently being designed and tested. Time-critical sections of the code are implemented in C.

The package also provides support for supervised discretization and mining Class Association Rules (CARs).

Installation

Stable CRAN version: install from within R with

install.packages("arulesCBA")

Current development version:

library("devtools")
install_github("ianjjohnson/arulesCBA")

Usage

library("arulesCBA")
data("iris")
 
# learn a classifier using automatic default discretization
classifier <- CBA(Species ~ ., data = iris, supp = 0.05, conf = 0.9)
classifier
 
  CBA Classifier Object
  Class: Species=setosa, Species=versicolor, Species=virginica
  Default Class: Species=setosa
  Number of rules: 8
  Classification method: first 
  Description: CBA algorithm by Liu, et al. 1998 with support=0.05 and confidence=0.9
 
# make predictions for the first few instances of iris
predict(classifier, head(iris))
 
   [1] setosa setosa setosa setosa setosa setosa
   Levels: setosa versicolor virginica

References

• Liu, B. Hsu, W. and Ma, Y (1998). Integrating Classification and Association Rule Mining. KDD'98 Proceedings of the Fourth International Conference on Knowledge Discovery and Data Mining, New York, 27-31 August. AAAI. pp. 80-86.