Association Rule Learning

Apriori

How it works

The association scenario can be movie or product recommendations: "People watched/bought A also watched/bought what.... B or C or something else?

s u p p o r t (M) = \frac{# u s e r w a t c h l i s t s c o n t a i n i n g M}{# u s e r w a t c h l i s t s}

c o n f i d e n c e (M_{1} - > M_{2}) = \frac{# u s e r w a t c h l i s t s c o n t a i n i n g M_{1} a n d M_{2}}{# u s e r w a t c h l i s t s c o n t a i n i n g M_{1}}

l i f t (M_{1} - > M_{2}) = \frac{c o n f i d e n c e (M_{1} - > M_{2})}{s u p p o r t (M_{1})}

set a minimum support and confidence
take all the subsets in transactions having higher support than minimum support
take all the association rules of these subjects having higher confidence than minimum confidence
sort the rules by decreasing lift

sklearn does not include Apriori modules!!! Use apyori instead.

Pros & Cons

pros
- results are intuitive and interpretatable
- exhaustive approach as it finds all rules based on the confidence and support
cons
- generates many uninteresting itemsets
- computationally and memory intensive
- results in many overlapping item sets

Eclat

Eclat is similar to Apriori but only uses support

How it works

set a minimum support
take all the subsets in transactions having higher support than minimum support
sort these subsets by decreasing support