An Obstacle Avoidance Strategy to Ant Colony Optimization Algorithm for Classification in Event Logs

This paper presents a novel approach to the ant colony optimization algorithm by using an obstacle avoidance strategy for mining classification rules from event log file datasets. An obstacle is purported to be present on a path as a classification rule is incrementally discovered if the rule convergence time is high or the degree of change between successive modifications to the rule exceeds a certain threshold value. By assigning zones to complete paths in a region based on the associated average obstacle density and prioritizing them, classification rules are discovered in descending order of the priorities of zones to enable faster mining in more obstacle-free paths. Experimental results are shown describing a comparative study with the popular C5 algorithm for the event log file datasets