I'm learning about association rules and came across the common interestingness measures support, confidence, lift and conviction.

I'm interested in the intuition behind your decision-making process while dealing with those measures. What I'm looking for is practical advice which I can apply during my data analysis projects.

For example, the second answer from Gericke Potgieter gives a good rule of thumb (high lift, low conviction) for determining the reliability of a discovered rule.

Some intuition which I commonly use:

higher support     --> rule applies to more records
higher confidence  --> chance that the rule is true for some record is higher
higher lift        --> chance that the rule is just a coincidence is lower
higher conviction  --> ?

What is your opinion about my intuition?

What are some of the heuristics/rules of thumb you use for making decisions regarding the above-mentioned measures?

  • $\begingroup$ Hi @anna, did you find the answer you are looking for? $\endgroup$ – 2943 Nov 9 '17 at 12:05
  • 1
    $\begingroup$ citeseerx.ist.psu.edu/viewdoc/… I hope this paper helps $\endgroup$ – 2943 Nov 9 '17 at 12:33
  • $\begingroup$ This paper called Selecting the right objective measure for association analysis cse.msu.edu/~ptan/papers/IS.pdf can help you, because he explains when you may use a measure like confidence or `lift. He also explains some intuition behind the measures with nice examples, and also the formulas and the properties. $\endgroup$ – igorkf Aug 20 '20 at 13:43

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.