DIAGNOSIS OF HEART DISEASE USING MIXED CLASSIFIER

Abstract

At present, there are many studies related to the use of medical information, in order to assist in the analysis and diagnosis of the patients. In the field of computer science, there are methods to assist in analysis and diagnosis of the various diseases using the data mining techniques. Data mining is a popular technique used for analyzing a large number of data, to find the relationship of the hidden information in those data and applied for the benefits of the organization. Classification technique based on the supervised learning is one of the data mining techniques, which divides the dataset into two subsets: training set and testing set. The training set is used to create a classification model and this classification model will be evaluated the performance by the testing set. In this thesis, three individual classifies, namely, Naïve Bayes approach, Support Vector Machine and K-Nearest Neighbor method are combined using an artificial neural network with backpropagation learning to enhance the performance of classification. This method is called a mixed classifier. The process of the mixed classifier divides the classification into two major phases. In the first phase, the heart disease dataset obtained from the UCI Machine Learning Repository is classified by Naïve Bayes approach, Support Vector Machine and K-Nearest Neighbor method. Then in the second phase, the results of individual classifiers become the input for classifying with the artificial neural network. The performance of the mixed classifier is measured by three heterogeneous measures: accuracy, False Positive Rate (FPR) and False Negative Rate (FNR). From the experimental results, the mixed classifier is more accurate and yields better FPR than the other classifiers.