Computational intelligence approach for prediction of breast cancer using particle swarm optimization: A comparative study of the results with reduced set of attributes

Abstract

Data mining is termed as mining of relevant information from abundant volumes of data in order to promote the business activity and decision making capability. The exploratory data analysis is a similar technique for summarizing and identifying the patterns in the data. This exploratory data analysis is a statistical model like a regression model, discriminant model or a clustering model, which is built from the data and is utilised for prediction, classification, or hypothesis verification. We presented a rule discovery algorithm based on swarm intelligence in order to identify the standard production rules and apply the rule pruning mechanism to shorten the rule. The proposed algorithm uses Wisconsin Breast Cancer and Mammographic Mass Data. This algorithm when applied on Wisconsin Breast Cancer found to be more accurate compared to the existing classification algorithms namely C4.5 and Classification using Regression Trees. A comparative study has been performed with dimensionality reduction using entropy based Information gain measure.