Support, Confidence, and Lift are the three fundamental metrics used in Market Basket Analysis to evaluate whether a product association is meaningful enough to act on.
Support Support measures how frequently a product combination appears across all transactions.
Formula: Support(A ∪ B) = Transactions containing both A and B ÷ Total transactions
A support threshold filters out rare combinations. Too high and you miss niche opportunities; too low and you get noise.
Confidence Confidence measures the reliability of a rule — given A is purchased, how likely is B?
Formula: Confidence(A → B) = Support(A ∪ B) ÷ Support(A)
High confidence (e.g., 0.7+) means the rule is dependable. But confidence alone can be misleading if product B is simply popular.
Lift Lift measures the true strength of an association by accounting for the baseline popularity of each product.
Formula: Lift(A → B) = Confidence(A → B) ÷ Support(B)
- Lift = 1: No association (products are purchased independently)
- Lift > 1: Positive association (products are purchased together more than expected)
- Lift < 1: Negative association (products are purchased together less than expected)
The golden rule: Focus on associations with high confidence AND lift significantly > 1. These represent genuine, actionable product relationships rather than statistical artifacts.
