In association rules, what does 'support' refer to?

Support in the context of association rules refers to the frequency of transactions containing a specific itemset. This metric is crucial because it helps identify how often items appear together in the dataset, thereby indicating the strength of the association between them. For example, if we have a dataset of grocery transactions, the support of the itemset {bread, butter} would represent the proportion of transactions that include both bread and butter. High support suggests that the items are frequently purchased together, which is valuable information for market basket analysis and other applications.

Understanding support is essential for evaluating the significance of a rule since it helps distinguish between findings that are meaningful and those that are merely coincidental. Support is not concerned with the reliability or complexity of a rule, nor does it account for the total count of individual items in a dataset. It specifically focuses on the co-occurrence of itemsets within transactions, making it a fundamental concept in data mining and association rule learning.