MapReduce based Classification for Microarray data using Parallel Genetic Algorithm

Abstract

Inorder to uncover thousands of genes Microarray   produces high throughput is used. Only few gene expression data out of thousands of data is used for disease predication and also for disease classification in medical environment.  To find such initial coexpressed gene groups of clusters whose joint expression is strongly related with the class label A Supervised attribute clustering is used. By sharing the information between each attributes the Mutual Information uses the information of sample varieties to measure the similarity among the attributes. From this the redundant and irrelevant attributes are removed. After forming the clusters the PGA is used to find the optimal feature and is given as mapper function so as to improve the class separability. Using this method the diagnosis can be made easier and effective since its done parallelly. The predictive accuracy is estimated using all the three classifiers such as K-nearest neighbours including naive bayes and Support Vector machine. Thus the overall approach used reducer function which provides excellent predictive capability for accurate medical diagnosis.