Publication

Abstract : In the modern state-of-art of technology, Machine Learning emerges out as a boom to extract information from mammoth dataset and transform into acquainted information. In particular, Clustering (Unsupervised learning) and Classification (Supervised learning) are the two predominant Machine Learning approaches emphasized here. However, data and constraints are known primarily in Classification, they are unknown in Clustering. In recent times, Clustering and Classification started playing significant role in the area of innumerable applications like Cognitive Services, Image Recognition and Manipulation, Business and Legal, Text and Language, Medical, Weather Forecast, Genetics, Bio-informatics and so on. A few recently established machine learning methodologies are depicted here, with a provision to convey vital concepts to classification and clustering experts. The aim of this paper is to focus various Machine Learning techniques through which one can predict the heart disease of a patient by analyzing various medical diagnostic parameters and patterns. A comparative study is made with respect to both unsupervised learning (Partitioning-based, Hierarchicalbased, Density-based and Model-based clustering) and supervised learning (SVM, Random Forest (RF), Decision tree (DT) and K-nn) empirically with the inclusion of large number of datasets. The results are explicit that Decision Tree has more classification accuracy of 73% thereby correlating K-means, K-modes, K-medoids, CLARANS, PAM, FCM, CLARA, DBSCAN, Ward's, ROCK, FCM, SVM, EM, OPTICS, Random Forest and K-nn. In this perspective, R X64 3.1.3 is used as a tool to determine the accuracy of aforementioned algorithms.