Discovering Positive Association of ASD Attributes With Class Using Multi Objective Cultural Algorithm

Association rule mining (ARM) is a common and most preferable research method for bringing out the fascinating relations between the variables provided in any data set. It brings out the knowledge by satisfying the user defined values and criteria measures specified by the researcher. Frequent item set generation is a well-known method carried out by many researchers to retrieve interesting correlations among the variables that helps in decision making. The accuracy of the brought out rules by ARM is good enough to provide a conclusion on research studies. This can be improved by incorporating optimization like heuristic search techniques. In this paper cultural algorithm is used to improve the performance of rule mining by optimization which is required for categorizing the risk level of ASD individuals. Optimization is utilized in health care domain for generating optimized rules to analyze the frequently combined attributes among the patient’s data. It gracefully improves the result finding process which will be tranquil to conclude the decision. The Cultural algorithm fit in to the larger course of evolutionary algorithms that is inspired by natural evolution. In this paper multi objective optimization technique is proposed by incorporating ARM and cultural algorithm by considering different objectives namely support, confidence, lift and completeness of the rule to find the positive association of Autism Spectrum Disorder (ASD) screening data features with positive class. The result of this research depicts the positive association with improved performance along with reduced number of rules.