Am trying to code an ACO algorithm in C#. Am having trouble understanding the pruning process.
I mean, after we create a Rule, we does the pruning of that Rule:
I have referred many other papers and saw that, in Pruning, we try to temporarily remove terms one by one and calculates the Quality and the term that improves the Quality upon removal, is permanently removed. And this shortened Rule is again pruned until there's no other for consideration:
As per that paper, the Quality calculation is based on this equation:
Am having doubt at those terms mentioned in that equation. So I researched and found another one which was kind of looking better:
So in the above equation, you will see the terms TP, FP, TN, FN. And having doubt regarding how we find out the cases covered by this rule and cases not covered by this rule
Is it like we try to count the number of rows in our validation set, that matches the terms in our Rule? Any ideas? Am really stuck at this part.
