✦ScienexySCIENCE KNOWLEDGE LIBRARY
Science reference · SCI-CS-003-090

Association Rule Learning

Association rule learning discovers relationships and co-occurrence patterns among items in transactional or categorical datasets.

Support(X → Y) = Support(X ∪ Y); Confidence(X → Y) = Support(X ∪ Y) / Support(X); Lift(X → Y) = Confidence(X → Y) / Support(Y)

Explanation

Association rule learning is an unsupervised data-mining technique used to discover relationships among items in large collections of transactions. A rule has the form X → Y, meaning that transactions containing itemset X are associated with itemset Y. Important measures include support, confidence, and lift. Support measures how frequently the combined itemsets occur in the dataset. Confidence measures the proportion of transactions containing X that also contain Y. Lift compares the observed co-occurrence of X and Y with the co-occurrence expected if they were independent. A lift greater than one indicates positive association, a lift equal to one indicates independence, and a lift below one indicates negative association under the standard definition. Algorithms such as Apriori and FP-Growth efficiently discover frequent itemsets from transactional data. Association rules describe statistical relationships and do not establish causation.

Variables

SymbolMeaning

X

Antecedent or itemset on the left side of a rule

Y

Consequent or itemset on the right side of a rule

Support(X)

Proportion of transactions containing itemset X

Support(X ∪ Y)

Proportion of transactions containing both X and Y

Confidence(X → Y)

Conditional frequency of Y among transactions containing X

Lift(X → Y)

Strength of association relative to the independence baseline

Example or application

A supermarket analyzes thousands of shopping baskets and discovers that customers who purchase bread frequently also purchase butter. The rule can help the retailer plan product placement or recommendations.

Text

Association rule learning identifies which items or events frequently occur together and expresses these relationships as rules.

Key Points

  • Association rules identify co-occurrence patterns.

  • A rule is written as X → Y.

  • Support measures how frequently an itemset occurs.

  • Confidence measures how often Y occurs when X occurs.

  • Lift compares observed association with expected association under independence.

  • Apriori and FP-Growth are common frequent-itemset mining algorithms.

  • Association does not prove causation.

Applications

  • Market basket analysis

  • Product recommendations

  • Retail inventory planning

  • Web usage analysis

  • Healthcare pattern discovery

Learning Outcomes

  • Explain association rule learning.

  • Interpret antecedents and consequents.

  • Calculate and distinguish support, confidence, and lift.

  • Understand how frequent-itemset algorithms discover rules.

Practice Questions

  • What is an association rule?

  • How does support differ from confidence?

  • What does a lift value greater than one indicate?

  • Why does an association rule not prove causation?

Related Concepts

  • Market Basket Analysis

  • Frequent Itemsets

  • Apriori Algorithm

  • FP-Growth

  • Support and Confidence

Common Mistakes

  • Treating high confidence as proof of a strong association without checking support and lift.

  • Assuming an association rule proves that one item causes another.

  • Ignoring the minimum support threshold.

  • Generating too many rules without considering their practical usefulness.

Sources and verification

Sources are shown as provided in the project data. Not every source has yet been independently verified. Please report errors or missing citations.

  • Agrawal, R., Imieliński, T., and Swami, A. (1993). Mining Association Rules between Sets of Items in Large Databases.
  • Han, J., Kamber, M., and Pei, J. (2011). Data Mining: Concepts and Techniques.

Suggest a correction or source · More in Artificial Intelligence & Machine Learning