PERFORMA ANALYSIS OF CLUSTERING OF THYROID DRUG DATA USING FUZZY AND M-CLUST
Katikireddy Srinivas, K. Kiran · Journal of Critical Reviews · 2020
To day we know the importance of machine learning at almost all areas from email spam filters to automatic chat bot application to specific business due to decision making is needed to support the various day to day activities of the business[19]. From this machine learning, I inspired to develop an automated clustering system for thyroid diagnosed patients to prescribe appropriate drugs from the available drug data set at various illness conditions[20].In this scenario I used computer science knowledge based on the physio chemical properties and enzyme inhibition properties of drugs provided by standard drug bank repository ie www.drugbank.ca and www.malacards.org . Here I applied existing clustering techniques along with hybrid combination of k-means, k-medoid, hierarchical methods and Fuzzy k-means [12]variants also to determine an appropriate set of drugs from the given dataset for different types of thyroid disease. In this paper I would like to compare the performance of Fuzzy Clustering and Model based clustering [1]using fanny(), m-clust(), Silhoutte measures[3].`Finally, the Clustered drugs are appropriate for Hyper thyroid, Hypo Thyroid and Normal cases based on the associated physio chemical and enzyme inhibition properties of each drug of the drug bank.