R package arules - Mining Association Rules and Frequent Itemsets

Introduction

The arules package family for R provides the infrastructure for representing, manipulating and analyzing transaction data and patterns using frequent itemsets and association rules. The package also provides a wide range of interest measures and mining algorithms including the code of Christian Borgelt’s popular and efficient C implementations of the association mining algorithms Apriori and Eclat. In addition, the following mining algorithms are available via fim4r:

• Apriori
• Eclat
• Carpenter
• FPgrowth
• IsTa
• RElim
• SaM

Code examples can be found in Chapter 5 of the web book R Companion for Introduction to Data Mining.

To cite package ‘arules’ in publications use:

Hahsler M, Gruen B, Hornik K (2005). “arules - A Computational Environment for Mining Association Rules and Frequent Item Sets.” Journal of Statistical Software, 14(15), 1-25. ISSN 1548-7660. doi:10.18637/jss.v014.i15 https://doi.org/10.18637/jss.v014.i15.

@Article{,
  title = {arules -- {A} Computational Environment for Mining Association Rules and Frequent Item Sets},
  author = {Michael Hahsler and Bettina Gruen and Kurt Hornik},
  year = {2005},
  journal = {Journal of Statistical Software},
  volume = {14},
  number = {15},
  pages = {1--25},
  doi = {10.18637/jss.v014.i15},
  month = {October},
  issn = {1548-7660},
}

Packages

arules core packages

• arules: arules base package with data structures, mining algorithms (APRIORI and ECLAT), interest measures.
• arulesViz: Visualization of association rules.
• arulesCBA: Classification algorithms based on association rules (includes CBA).
• arulesSequences: Mining frequent sequences (cSPADE).

Other related packages

Additional mining algorithms

• arulesNBMiner: Mining NB-frequent itemsets and NB-precise rules.
• fim4r: Provides fast implementations for several mining algorithms. An interface function called fim4r() is provided in arules.
• opusminer: OPUS Miner algorithm for finding the top-k productive, non-redundant itemsets. Call opus() with format = 'itemsets'.

In-database analytics

• ibmdbR: IBM in-database analytics for R can calculate association rules from a database table.

Interface

• rattle: Provides a graphical user interface for association rule mining.
• pmml: Generates PMML (predictive model markup language) for association rules.

Classification

• arc: Alternative CBA implementation.
• rCBA: Alternative CBA implementation.
• qCBA: Quantitative Classification by Association Rules.
• sblr: Scalable Bayesian rule lists algorithm for classification.

Recommendation/Prediction

• recommenderlab: Supports creating predictions using association rules.

The following R packages use arules: arc, arlclustering, arulesCBA, arulesNBMiner, arulesSequences, arulesViz, clickstream, CLONETv2, CRE, ctsem, daltoolbox, discnorm, fcaR, fdm2id, gp3sequences, GroupBN, ibmdbR, icdhelper, manydist, nuggets, opusminer, pervasive, pmml, qCBA, RareComb, rattle, rCBA, recommenderlab, rgnoisefilt, RulesTools, sbrl, SurvivalTests, TELP, tidylearn

Installation

Stable CRAN version: Install from within R with

install.packages("arules")

Current development version: Install from r-universe.

install.packages("arules",
    repos = c("https://mhahsler.r-universe.dev",
              "https://cloud.r-project.org/"))

Usage

Load package and mine some association rules.

library("arules")
data("IncomeESL")

trans <- transactions(IncomeESL)
trans

## transactions in sparse format with
##  8993 transactions (rows) and
##  84 items (columns)

rules <- apriori(trans, supp = 0.1, conf = 0.9, target = "rules")

## Apriori
## 
## Parameter specification:
##  confidence minval smax arem  aval originalSupport maxtime support minlen
##         0.9    0.1    1 none FALSE            TRUE       5     0.1      1
##  maxlen target  ext
##      10  rules TRUE
## 
## Algorithmic control:
##  filter tree heap memopt load sort verbose
##     0.1 TRUE TRUE  FALSE TRUE    2    TRUE
## 
## Absolute minimum support count: 899 
## 
## set item appearances ...[0 item(s)] done [0.00s].
## set transactions ...[84 item(s), 8993 transaction(s)] done [0.01s].
## sorting and recoding items ... [42 item(s)] done [0.00s].
## creating transaction tree ... done [0.00s].
## checking subsets of size 1 2 3 4 5 6 done [0.02s].
## writing ... [457 rule(s)] done [0.00s].
## creating S4 object  ... done [0.00s].

Inspect the rules with the highest lift.

inspect(head(rules, n = 3, by = "lift"))

##     lhs                           rhs                      support confidence coverage lift count
## [1] {dual incomes=no,                                                                            
##      householder status=own}   => {marital status=married}    0.10       0.97     0.10  2.6   914
## [2] {years in bay area=>10,                                                                      
##      dual incomes=yes,                                                                           
##      type of home=house}       => {marital status=married}    0.10       0.96     0.10  2.6   902
## [3] {dual incomes=yes,                                                                           
##      householder status=own,                                                                     
##      type of home=house,                                                                         
##      language in home=english} => {marital status=married}    0.11       0.96     0.11  2.6   988

The following short guides cover the main workflows:

• Getting started
• Preparing transaction data
• Mining and pruning rules
• Interest measures
• Item hierarchies

The theoretical background and the original description of the package are available in the paper Introduction to arules.

Using arules with tidyverse

arules works seamlessly with tidyverse. For example:

• dplyr can be used for cleaning and preparing the transactions.
• transactions() and other functions accept tibble as input.
• Functions in arules can be connected with the pipe operator |>.
• arulesViz provides visualizations based on ggplot2.

For example, we can remove the ethnic information column before creating transactions and then mine and inspect rules.

library("tidyverse")
library("arules")
data("IncomeESL")

trans <- IncomeESL |>
  select(-`ethnic classification`) |>
  transactions()
rules <- trans |>
  apriori(
    supp = 0.1, conf = 0.9, target = "rules",
    control = list(verbose = FALSE)
  )
rules |>
  head(3, by = "lift") |>
  as("data.frame") |>
  tibble()

## # A tibble: 3 × 6
##   rules                                  support confidence coverage  lift count
##   <chr>                                    <dbl>      <dbl>    <dbl> <dbl> <int>
## 1 {dual incomes=no,householder status=o…   0.102      0.971    0.105  2.62   914
## 2 {years in bay area=>10,dual incomes=y…   0.100      0.961    0.104  2.59   902
## 3 {dual incomes=yes,householder status=…   0.110      0.960    0.114  2.59   988

Using arules from Python

arules and arulesViz can now be used directly from Python with the Python package arulespy available from PyPI.

Support

Please report bugs here on GitHub. Questions should be posted on stackoverflow and tagged with arules.

References

• Michael Hahsler. ARULESPY: Exploring association rules and frequent itemsets in Python. arXiv:2305.15263 [cs.DB], May 2023.
• Michael Hahsler. An R Companion for Introduction to Data Mining: Chapter 5, 2021, URL: https://mhahsler.github.io/Introduction_to_Data_Mining_R_Examples/book/
• Hahsler, Michael. A Probabilistic Comparison of Commonly Used Interest Measures for Association Rules, 2015, URL: https://mhahsler.github.io/arules/docs/measures.
• Michael Hahsler, Sudheer Chelluboina, Kurt Hornik, and Christian Buchta. The arules R-package ecosystem: Analyzing interesting patterns from large transaction datasets. Journal of Machine Learning Research, 12:1977-1981, 2011.
• Michael Hahsler, Bettina Grün and Kurt Hornik. arules - A Computational Environment for Mining Association Rules and Frequent Item Sets. Journal of Statistical Software, 14(15), 2005.