ArulesCBA Versions Save

Classification Based on Association Rules in R

arulesCBA_1.2.1

2 years ago

arulesCBA 1.2.1 (2021-11-20)

  • mineCARs now uses by default minimum LHS-support (via parameter originalSupport = FALSE).
  • the CBA_ruleset function is now used consistently as the constructor for CBA objects.
  • added transactionCoverage.
  • added uncoveredClassExamples.
  • added uncoveredMajorityClass.
  • added transactions2DF to convert transactions to a data.frame.
  • RCAR is now faster (does not run glmnet again for the chosen lambda) and returns the whole regularization path.
  • prepareTransactions now automatically add a negative class item if needed.
  • moved the experimental algorithms wCBA and bCBA to Work.
  • R/Weka-based classifiers have now a default class.

arulesCBA_1.2.0

4 years ago
  • Version 1.2.0 has a major interface cleanup. This might require some change in existing code.
  • The classifiers now use as the default a min. confidence of .5 and maxlen of 5 (max. rule length).
  • CBA now includes a default rule in the rule base.
  • added prepareTransactions to dicretize and convert a data.frame into transactions.
  • added new function response to convert class items to class labels (factors).
  • added new function majorityClass.
  • added FOIL.
  • added RIPPER C4.5, and PART (via RWeka).
  • added PRM, CPAR and CMAR (via LUCS-KDD Software Library).
  • added datasets Mushroom and Lymphography.