Adding mutual information metric to the association_rules() function

[kristianperriu]

Mutual Information is highly useful in Market Basket Analysis because it measures the statistical dependence between items, helping to identify patterns of association beyond simple frequency counts. Helps determine how much knowing the presence of one product in a transaction reduces uncertainty about the presence of another product. This is especially beneficial for identifying non-obvious associations between items that might not appear frequently together but are still highly dependent.

Since this function is highly used in the context of Market Basket Analysis it could be useful to provide another layer of information to help the users reduce the number of possibly redundant or useless rules after generating the rules from an algorithm such as Apriori or FP-Growth