Question

我有从python3的pyfpgrowth包中获得的关联规则。

[In] AssociationRules = pyfpgrowth.generate_association_rules(FrequentPattern, 0.8)

[In] AssociationRules

[Out] 
{('Longitude 22', 'Magnitude 97'): (('Depth 108', 'Latitude 77'), 1.0),
('Latitude 123',): (('Longitude 5', 'Region Name 20'), 0.8),
('Latitude 147',): (('Longitude 36', 'Region Name 57'), 0.8),
('Longitude 36',): (('Latitude 147', 'Region Name 57'), 0.8),
('Longitude 41', 'Region Name 16'): (('Latitude 149',), 1.0),
('Longitude 23', 'Magnitude 39'): (('Region Name 46',), 1.0),
('Region Name 20',): (('Longitude 5',), 0.8571428571428571),
('Depth 17', 'Longitude 23'): (('Region Name 46',), 1.0),
('Latitude 119', 'Longitude 17'): (('Region Name 37',), 0.8),
('Depth 17', 'Latitude 119', 'Longitude 17'): (('Region Name 37',), 1.0),
('Longitude 4',): (('Region Name 35',), 0.9),
('Longitude 4', 'Region Name 35'): (('Latitude 187',), 1.0),
('Depth 17', 'Longitude 4'): (('Region Name 35',), 1.0),
('Depth 17', 'Longitude 4', 'Region Name 35'): (('Latitude 187',), 1.0)}

根据协会规则，我尝试使用以下代码查找提升值：

def lift_measure(FreqItems, N):
    lift = []
    cnt = 0

    for items, support in FreqItems.items():
        if(len(items) > 1):
            for i in range(1,len(items)):
                lst = list(items)
                antecedent = lst[:len(lst) - i]
                consequent = lst[-i:]

                conf =  float(FreqItems[frozenset(items)]/FreqItems[frozenset(antecedent)])
                if (conf >= confidence):
                    hasil = float((FreqItems[frozenset(items)]/N)/
                            ((FreqItems[frozenset(antecedent)]/N)* 
                            (FreqItems[frozenset(consequent)]/N)))

                    lift.append((antecedent,consequent,support,conf,hasil))

    return lift

之后，我运行该函数：

[In] lift_measure(AssociationRules, 0.2)

然后我得到这样的错误：

[Out] frozenset({'Longitude 22', 'Magnitude 97'})

非常感谢！

使用Lift从FP-Growth评估关联规则

0 个答案: