Welcome to ARMxtend's documentation!

ARMxtend (association rule mining extensions) is a Python library for mining association rules, covering the crisp and fuzzy cases, on a single machine as well as on Apache Spark for Big Data and streaming scenarios, plus meta-association rules for summarizing rules found across multiple datasets.

DOI Python 3.6 Python 3.7

Features

Crisp Fuzzy (alpha-cuts)
Single machine ARM.association_rules, FFIM.fpgrowth FIM.FARE, FFIM.fuzzy_fpgrowth
Big Data (Spark) FIM.apriori (DApriori/DAprioriTID), FIM.Eclat.DECLAT, FIM.BD_ARE FIM.BD_FARE.FuzzyDAprioriTID, FIM.Eclat.FuzzyDECLAT
Streaming (Spark) SFIM (FIMoTS) + SARE.extractAssociationRules -
Meta-rules (rules about rules, across multiple datasets) ARM.meta_rules.crisp_meta_association_rules ARM.meta_rules.fuzzy_meta_association_rules

Plus preprocessing.FuzzyLib for fuzzifying numeric attributes and VizARM.AREtoGraph for transforming rules into the VizARE typed item/rule graph (JGF/GraphML/DOT).

See the Quick Start for runnable examples of every one of these, and the User Guide for the full reference of each module.

Citing

If you use ARMxtend as part of your workflow in a scientific publication, please consider citing the underlying research:

DOI

@article{fernandez2019finding,
  title={Finding tendencies in streaming data using big data frequent itemset mining},
  author={Fernandez-Basso, Carlos and Francisco-Agra, Abel J and Martin-Bautista, Maria J and Ruiz, M Dolores},
  journal={Knowledge-Based Systems},
  volume={163},
  pages={666--674},
  year={2019},
  publisher={Elsevier}
}

See Citing ARMxtend for the references behind each individual module (crisp/fuzzy Big Data algorithms, fuzzy association rules, meta-association rules).