FIM.BD_FARE -- FuzzyDAprioriTID (Spark)

Requires the spark extra (pip install -e ".[spark]", see Installation).

Fuzzy frequent itemset mining on Apache Spark, inspired by Apriori-TID, implementing Algorithm 1 ("BDFARE-Apriori"/"BDFARE-Apriori-TID") of Fernandez-Basso, Ruiz & Martin-Bautista (2021) -- see Citing ARMxtend.

FuzzyDAprioriTID.run(sc, transactions, min_supp, num_alpha)

• sc (pyspark.SparkContext).
• transactions (RDD[Iterable[Tuple[str, float]]]): each transaction is a list of (item, membership degree in [0, 1]) pairs.
• min_supp (float): minimum aggregated fuzzy support (FSupp), in (0, 1].
• num_alpha (int): number of alpha-cuts used to decompose the fuzzy database (the paper experimentally shows 10 equidistributed alpha-cuts are enough to uncover every significant fuzzy association rule).

Returns dict {itemset_key: numpy.ndarray(num_alpha)}: the bit-list of relative support of each frequent itemset at every alpha-cut (see FIM._shared.alpha_cuts) -- the same output convention as FIM.Eclat.FuzzyDECLAT and FFIM.fuzzy_fpgrowth, so all three are interchangeable as input to FIM.FARE.fuzzy_association_rules.

from ARMxtend.FIM.BD_FARE import FuzzyDAprioriTID

fuzzy_transactions = sc.parallelize([
    [("cold", 1.0), ("low_humidity", 0.8)],
    [("cold", 0.9), ("low_humidity", 0.6)],
    [("warm", 0.7), ("low_humidity", 0.9)],
])
freq_itemsets = FuzzyDAprioriTID.run(sc, fuzzy_transactions, min_supp=0.3, num_alpha=10)