"No one is harder on a talented person than the person themselves" - Linda Wilkinson ; "Trust your guts and don't follow the herd" ; "Validate direction not destination" ;

October 23, 2020

Good Read - Paper - Frequent Item-set Mining without Ubiquitous Items

 Link 

We use Apriori to find Frequently occurring items.

Example demonstration
  • No of Transactions with A - 10
  • No of Transactions with B - 50
  • No of Transactions with A+B = 5
  • Total Transactions - 100
Sup(A) = 10/100 = 0.1
Conf(A-B) = 5/10 = 1/2 = 0.5
Lift = 0.5/0.1 = 5
A->B - > 1, Likelihood

Sup (B) = 50/100 = 1/2 = 0.5
Conf(B-A) = 5/50 = 1/10 = 0.1
Lift = 0.1/0.5 = 1/5 = 0.2
B->A - 0.2 < 1 Less Likelihood

A variation of it without frequently occurring items. The ubiquitousness threshold is similar to the support threshold but, it filters out the items with a frequency higher than the threshold.



Keep Thinking!!!

No comments: