AKAMAS: Mining association rules using a new algorithm for the assessment of the risk of coronary heart events
Joseph A. Moutiris, M. Karaolis, L. Papaconstantinou, Constantinos S. Pattichis · 2009
Many algorithms have been developed for rule mining in large transaction databases. Discovery of some important association rules is a main database mining problem. The objective of this study was to develop a new data mining algorithm named AKAMAS using different measures to extract the most important association rules for the assessment of heart event related risk factors. The implemented measures were: support, confidence, p-value, chi square, coverage, prevalence, recall, specificity, accuracy, lift, leverage, added value, relative risk, odds ratio, and conviction. The AKAMAS algorithm is a variant of the Apriori algorithm, the main difference is that it does not use the iterative technique of k-itemset to build the (k +1)- itemsets. It needs only one pass for extracting frequent itemsets. Although AKAMAS gave similar rules to Apriori it offers a wide selection of measures for filtering the best rules, including the computation of the chi square test, and its associated probability value (i.e. if a rule is statistically significant or not). Moreover, the rules are more comprehensively represented and are more easily to interpret.