The Comparative Analysis of Machine Learning Techniques for Gestational Diabetics Prediction

Abstract

Diabetes became the leading disease to humanity from youth age to old age. There is a fast increment in the development of diabetic patients because of different causes, for example, bacterial or viral contamination, poisonous or substance blend in with the nourishment, autoimmune response, obesity, change in ways of life, dietary pattern, environmental pollution and so on. Numerous complications happen if diabetes stays untreated and unidentified. Consequently, a diabetes diagnosis is exceptionally pivotal to spare human life from diabetes. Data Analytics is a procedure of looking at and recognizing the concealed examples from a considerable measure of information to conclude. In Medical consideration, this diagnostic procedure can be completed utilizing AI algorithms for analyzing clinical information to construct the AI models to make clinical diagnoses. The thought process of this paper is to investigate the ways to deal with improve the exactness in the forecast of diabetes utilizing clinical information with different AI calculations and strategies. Hence, three AI classifications paradigms, in particular, Naïve Bayes', SVM, and Random Forest, are utilized in this literature for the recognition of diabetes at beginning times. Experiments performed on the Pima Indians Diabetes Database (PIDD), which is begun from the UCI AI repository. The performances of all three paradigms are assessed on different estimates like Precision, Accuracy, Support, F1-score, and Recall. Accuracy is estimated over accurately and inaccurately classified instances. Through our experimentation, we noticed that random forest algorithm outperforms with high accuracy when contrasted with different algorithms for the prediction of gestational diabetes.