Cohesion-Based Space-Time Optimized Framework for Confident Association Mining in Microarray Data

Numerous confident and interesting associations in datasets from diverse domains remain undiscovered for the constraint of classical minimal frequency. A lower choice for threshold frequency not only incurs huge cost of pattern explosion but also cuts reliability of discovered knowledge. Goal of the present paper is to devise a new framework addressing two necessities. First is discovery of confident associations that are not constrained by classical minimal frequency. Second is ensuring quality of the discovered rules as higher confidence does not necessarily imply better knowledge discovery. We propose a new property among items, terming it cohesion, and develop a cohesion-based scalable algorithm for pattern discovery. In addition to confidences of discovered associations, we compute average lift and DIR measure to assess their quality. Experiments with real and synthetic datasets as well as with microarray dataset prove superiority of the technique.