Association Rules

Association rule learning is a machine learning model that seeks to unearth the hidden patterns (in other words, relationships) in transaction data that describe the shopping habits of the customers of any retailer. The definition of an association rule was hinted at when the common probabilistic metrics were defined and explained previously.

Consider the imaginary frequent item set {Milk, Bread}. Two association rules can be formed from that item set: Milk Bread and Bread Milk. For simplicity, the first item set in the association rule is referred to as the antecedent, while the second item set in the association rule is referred to as the consequent. Once the association rules have been identified, all the previously discussed metrics can be computed to evaluate the validity of the association rules determining whether or not the rules can be leveraged in the decision-making process.

The establishment of an association rule is based on support and confidence. Support...