Installing ARMxtend

---

ARMxtend is not yet published on PyPI. Install it directly from a local checkout of the GitHub repository:

git clone https://github.com/cjferba/ARMxtend.git
cd ARMxtend
pip install -e .

This installs the core, single-machine part of the library: ARM (crisp association rules and meta-rules), FIM.FARE (fuzzy association rules), FFIM (crisp and fuzzy FP-Growth), VizARM and preprocessing. Its only dependencies are numpy, pandas and networkx.

Big Data / streaming extra

The Big Data and streaming algorithms (FIM.apriori, FIM.Eclat, FIM.BD_ARE, FIM.BD_FARE, SARE, SFIM) run on Apache Spark and additionally require pyspark. Install it with the spark extra:

pip install -e ".[spark]"

pyspark itself requires a Java runtime (JDK 8/11/17) to be installed and on your PATH; see the PySpark installation guide if pyspark fails to start a SparkContext.

Running the tests

pip install pytest
pytest tests/ ARMxtend/FIM/BD_ARE/TestFunctions.py

The tests covering the Spark-based algorithms run against tests/spark_stub.py, a local double of the small part of the Spark RDD/broadcast API these algorithms use, so they exercise the exact same production code without requiring a real cluster or a JVM.