Association of Data Mining and Biomedical Domain: Choosing the Best Algorithm for Analysis

Abstract

Data mining has been used prosperously in the favourably perceived areas, such as e-business, marketing, health domain and retail, because of which it is now applicable in knowledge discovery in databases (KDD). While data mining has become a much-lauded tool in medical-related fields, its role in the healthcare arena is still being explored. Data mining is mainly gaining its importance and usage in the areas of medicine and public health. Currently, most applications of data mining in healthcare can be categorised into two areas: decision support for clinical practice and decision making. In this paper, we studied and analysed the performance of the popular clustering methods like k-mean, Self Organized Maps and EM methods by using the bio medical data extracted from UCI Machine leaning repository. We made a comparative analysis between Entropy based mean (EBM) clustering approach and traditional clustering methods through accuracy, return on investment, reliability, time complexity and fitness. With this paper, we discussed the various issues to explore the strength of the data mining methodology.