Data Mining and Knowledge Discovery: Core Concepts
Multilevel Association Rules
Hierarchy Matters: Items in databases are structured from general to specific (e.g., Food → Dairy → Milk → Amul Milk).
Low-Level Items: Rarely appear, so have low support.
Support & Confidence: Support decreases as we get specific; confidence usually stays stable in the hierarchy.
Approaches: Uniform Minimum Support uses the same support for all levels, but may miss specific patterns. Reduced Minimum Support assigns different supports; higher levels get a higher threshold, while lower levels get a smaller threshold.
Search Strategies: Independent Search mines levels separately. Level-Cross Filtering explores child nodes only if the parent is frequent. Controlled Level-Cross Filtering balances exploration and... Continue reading "Data Mining and Knowledge Discovery: Core Concepts" »
English with a size of 1.43 MB