# Confidence vs. Count in association rule mining: which one is better?

I am writing a program that mines association rules from a large data set. I have an array of association rules, and I have to decide which ones are more representative of the patterns I am studying.

Example given:

[      rule        support   confidence   lift  count]

{burger -> beer}    0.125        1          8     2
{cheese -> coke}    0.25        0.25        1     4


Being "count" the number of times this rule is satisfied in the data set. The question is: which one should I consider as the most representative of the general behavior of my customers?

Is it better to choose a rule that has been observed a few times (low support and count), but it is very likely to happen (high confidence and lift)? Or is it better to choose a very repeated rule without such a high probability?