Parallelized computational methods for improved feature selection and classification of cancer types

Abstract

newlineData Mining was principally used to extract unknown knowledge from huge hidden data either stored in structured or unstructured format Data mining techniques find applications in wide variety of areas like healthcare stock market exchange huge transactional databases fraud detection weather forecasting disaster prediction retail marketing and many such fields This research focus is to exploit data mining and machine learning algorithms in healthcare application to predict cancer types and sub-types As clinical data is prone to grow exponentially applying data mining task on such huge data involves huge man power resources time and storage To conquer this issue parallel data mining algorithms were formulated which utilizes computing resources of many cores in a multicore processor as well as computing and storage resources of cluster of computers Data mining algorithms were executed in parallel on same dataset or on different dataset across different cores nodes in a cluster and provide results in very less time The stumbling block of this approach is that it reduces the efficiency accuracy of computational methods.