Association rules are used to find the relationships between the objects which are frequently used together. Association rules are generated by analyzing data for frequently occurring if/then patterns. Two basic criteria are used to evaluate them: support and confidence. An association rule is usually required to satisfy both a user-specified minimum support and a user-specified minimum confidence.
Measures
Support measures how often items in and occur together, as a proportion of all transactions:
Here, is the number of transactions containing both and , and is the total number of transactions.
Confidence measures how often items in occur in transactions that contain :
Here, is the number of transactions containing .
The rule means that when occurs, tends to occur as well. Such rules are commonly used in market basket analysis to identify relationships between product purchases. Confidence is the conditional frequency of given ; it does not, by itself, establish causation.
Related algorithms
References
- Kumbhare, T. A., & Chobe, S. V. (2014). An overview of association rule mining algorithms. *International Journal of Computer Science and Information Technologies, 5*(1), 927–930.