Knowledge discovery using Domain-Concept Mining approach for the Behavioral Risk Factor Surveillance System (BRFSS) data.
Wannapa Kay Mahamaneerat, Chi‐Ren Shyu · PubMed · 2006
The publicly available Behavioral Risk Factor Surveillance System (BRFSS) data is the largest telephone survey data set in the world. Often times, the data set is under-utilized due to its size and the difficulties to comprehend and explore the relationships among variables. With a traditional data mining approach, such as association rule (AR) mining, it is still not possible to discover valuable information under the existing computational power. To promote the usefulness of this rich data set efficiently, we propose a novel data mining approach called Domain-Concept Mining (DCM) that partitions data into groups of relevant domain-concept, then extracts associations among variables from each partition. The findings from the DCM show that it can efficiently discover relevant information from the BRFSS with respect to the previously published literature.