Apriori Algorithm: Definition of Lift and its interpretation

[Kansiri Apinantanakul]

Dear professors and all,

I’m writing this post to share my curiosity regarding the definition of “Lift” which is one result obtained from the Apriori Algorithm in R using arules package.

As professors mentioned that the lift may indicate how “interesting” of the result.
lift = confidence / expected_confidence.

I’m curious about the following topics.
1) Why the expected confidence came? Is it the same with support?
2) How could we interpret the “Lift”
I saw one of interpretation which is
A lift value greater than 1 indicates that the rule body and the rule head appear more often together than expected, this means that the occurrence of the rule body has a positive effect on the occurrence of the rule head.
A lift smaller than 1 indicates that the rule body and the rule head appear less often together than expected, this means that the occurrence of the rule body has a negative effect on the occurrence of the rule head.
A lift value near 1 indicates that the rule body and the rule head appear almost as often together as expected, this means that the occurrence of the rule body has almost no effect on the occurrence of the rule head.
(ref: https://www.ibm.com/docs/en/db2/10.5?topic=SSEPGG_10.5.0/com.ibm.im.model.doc/c_lift_in_an_association_rule.html)
Is this also correct ka?

Thank you in advance for your kind support ka.

[Anawat ratchatorn]

I also curious the same issue as Kansiri.

As I know from searching in the internet.

I found the formula to calculate lift as this picture

Formula of Lift (ref: https://bigdata.go.th/big-data-101/data-science/what-is-association-rule/)

From the formula, in my own understanding.
If lift of A -> B is high, it means that we won’t see A or B frequently among all data. but we see A and B frequently. That makes high lift value when we calculate it. That’s why higher lift value, more interesting the rule is.

For interpretation.
If lift > 1: positive correlation.
If lift = 1: A and B might be independent.
And if lift < 1: Negative Correlation.

I hope this can help. And please correct me if I understand incorrectly.

[Napisa Freya Sawamiphak]

I understood it as same as Anawat. In addition, I would like to suggest this site for self-learning. https://www.kdnuggets.com/2016/04/association-rules-apriori-algorithm-tutorial.html

I think this site provides content that is easy to understand. Hope it helps and pleases correct me if I misunderstood too!!

[Kansiri Apinantanakul]

Thank you for your sharing 🙂

[Author]

Posts