A Simplified Swarm Optimization for Discovering the Classification Rule Using Microarray Data of Breast Cancer

Microarray data analysis is a major line of research in bioinformatics. A significant trend in bioinformatics is identifying genes or gene groups that differentiate diseased tissues. Classification is necessary to make microarray data useful for application in medicine, and in related research such as disease diagnosis. Classification models have been developed using statistical methods such as logistic and multi-normal regression for data mining. However, the complexities of real-world classification problems, such as those in the medical domain, are highly dimensional. General statistical methods are inadequate for these complex problems. This study proposes simplified swarm optimization (SSO), an efficient methodology for discovering breast cancer classification rules. The data set was derived from the Stanford microarray database. The proposed approach enables simultaneous feature selection and pattern recognition. Experimental results indicate that SSO outperforms general data mining methods such as decision tree, neural network, support vector machine, etc. The proposed approach has potential applications in hospital decision-making and research such as predictive medicine.