Running Apriori

And finally, this is how we use the algorithm with our toy example:

let transactions = [["
", "
", "
", "
"], 
                    ["
", "
", "
"], 
                    ["
", "
", "
"], 
                    ["
", "
"], 
                    ["
", "
"], 
                    ["
", "
"], 
                    ["
"] 
] 
 
let apriori = Apriori<String>(transactions: transactions) 
let rules = apriori.associationRules(minSupport: 0.3, minConfidence: 0.5) 
for rule in rules { 
    print(rule) 
    print("Confidence: ", apriori.confidence(rule), "Lift: ", apriori.lift(rule), "Conviction: ", apriori.conviction(rule)) 
} 

It produces the following:

{ 
 → 
} 
Confidence:  0.8 Lift:  1.4 Conviction:  2.14285714285714 
{ 
 → 
} 
Confidence:  1.0 Lift:  1.4 Conviction:  inf 
{ 
 → 
} 
Confidence:  0.75 Lift:  1.3125 Conviction:  1.71428571428571 
{ 
 → 
} 
Confidence:  0.75 Lift:  1.3125 Conviction:  1.71428571428571 

Let's analyze what's going on here. The second rule has the maximum confidence as well as conviction...