Auto Determining Parameters in Class-Association Mining
Viet Phan-Luong · 2012
This work proposes an approach to determine automatically parameters in classification rule mining based on association rules. Such parameters are the thresholds of support and confidence, and the maximal size of rules. The approach is based on statistical data get on the dataset during the mining process. In particular, the thresholds of support and confidence are not fixed, but varied dependently on each other and on the size of rules.