Association Rules Selection Approach Based on Interesting Measures

In recent years, the size of data collection becomes growing significantly. The knowledge results in term of association rules obtained from the set of data are numerous and difficult to select. This paper proposed the approach to select the interesting subsets of association rules from the big association results. The selection criterion is based on wellknown interesting measures that are confidence, coverage, leverage and lift. The original association rules for Thai stock market in the period of April 10, 2013 to September 5, 2014 are investigated and applied to the selection approach. The selection approach reduces 246,084 association rules into 10 rules.