Evolving Temporal Association Rules with Genetic Algorithms

A novel framework for mining temporal association rules by discovering
itemsets with a genetic algorithm is introduced. Metaheuristics have been applied
to association rule mining, we show the efficacy of extending this to another
variant - temporal association rule mining. Our framework is an enhancement to
existing temporal association rule mining methods as it employs a genetic algorithm
to simultaneously search the rule space and temporal space. A methodology for
validating the ability of the proposed framework isolates target temporal itemsets in
synthetic datasets. The Iterative Rule Learning method successfully discovers these
targets in datasets with varying levels of difficulty.