Online ISSN: 2515-8260

Keywords : Confusion Matrix

Analyzing Diabetic Data Using Naive-Bayes Classifier

A. Sharmila Agnal; E. Saraswathi

European Journal of Molecular & Clinical Medicine, 2020, Volume 7, Issue 4, Pages 2687-2699

Approximately 422 million people across the world have diabetes, particularly in countries where the average income is in the middle and lower end of the economic spectrum. Statistics reveal that every year, about 1.6 million deaths are recorded which can be directly attributed to diabetes. The graph suggests that number of cases as well as the prevalence of diabetes have been steadily incrementing over the past few decades. Through this new implementation of the Bayesian Classifier, raw medical data is analyzed and the risk of diabetes diagnosis based on each patient’s medical information can be calculated. The raw data is converted into class labels and the likelihood of a positive potential diabetes case is derived, as a probability (≤1). This can not only be used by healthcare professionals but also by common users, and can be useful in detecting the risk and preventing it in time without taking any medical tests. This classifier uses very basic information that would be known to each patient or can easily be obtained.