Association Rules are if-then statements generated from transactional data that describe the probability of products being purchased together. They are the core output of Market Basket Analysis and take the form: "If A, then B" (written as A → B).
Example: {Cap} → {T-Shirt} with 68% confidence means that 68% of transactions containing a Cap also contain a T-Shirt.
Three key metrics define each rule:
- 1Support: The proportion of all transactions that contain both A and B. High support means the combination is common.
- 2Confidence: The proportion of transactions containing A that also contain B. High confidence means the rule is reliable.
- 3Lift: The ratio of observed support to expected support if A and B were independent. Lift > 1 indicates a genuine association (not just popularity).
Practical considerations:
Not all rules are equally valuable. A rule with high confidence but low lift might just reflect that the consequent (B) is an extremely popular product that appears in most orders. The best rules have high confidence AND high lift — indicating a genuine, actionable relationship rather than a statistical coincidence.
In e-commerce, the most valuable association rules are those you can act on: create bundles, configure cross-sell recommendations, or design targeted email campaigns around specific product affinities.
