关联规则Apriori

一个测试数据集

import pandas as pd

dataSet = [['A', 'C', 'D'],
           ['B', 'C', 'E'],
           ['A', 'B', 'C', 'E'],
           ['B', 'E']]

1、生成候选的1-项集

import itertools

def createC1(dataSet):  # 'C1' for Candidate-itemset of 1 item.
    # Flatten the dataSet, leave unique item
    C1 = set(itertools.chain(*dataSet))  
    
    # Transform to a list of frozenset
    return [frozenset([i]) for i in C1]

• 测试

C1 = createC1(dataSet)
list(C1)[:5]

[frozenset({'E'}),
 frozenset({'D'}),
 frozenset({'C'}),
 frozenset({'A'}),
 frozenset({'B'})]

2、按条件过滤，生成频繁1-项集

def scanD(dataSet, Ck, min_support): # 'Ck' for Candidate-set of k items.
    support = {
   }
    
    # Calculate the support of all itemsets
    for i in dataSet:
        for j in Ck:
            if j.issubset(i):
                support[j] = support.get(j, 0) + 1

    n = len(dataSet)
    
    # Return litemset with support
    return {
   k: v/n for k, v in support.items() if v/n >= min_support}

• 测试

min_support = 0.4
L1 = scanD(d