Prediction of Diabetes Using Hidden Naïve Bayes: Comparative Study

Abstract

Classification techniques performance varies widely with the techniques and the datasets employed. A process performance classifier lies in how accurately it categorizes the item. The technique of classification finds the relationships between the value of the predictor and the values of the goal. This paper is an in-depth analysis study of the classification of algorithms in data mining field for the hidden Naïve Bayes (HNB) classifier compared to state-of-the-art medical classifiers which have demonstrated HNB performance and the ability to increase prediction accuracy. This study examines the overall performance of the four machine learning techniques strategies on the diabetes dataset, including HNB, decision tree (DT) C4.5, Naive Bayes (NB), and support vector machine (SVM), to identify the possibility of creating predictive models with real impact. The classification techniques are studied and analyzed; thus, their effectiveness is tested for the Pima Indian Diabetes dataset in terms of accuracy, precision, F-measure, and recall, besides other performance measures.