Diabetes Detection Using Machine Learning Techniques

Abstract

Diabetes Mellitus is one of our country's significant public health issues. It is a metabolic disorder that has impacted thousands of people and is caused by excessive amounts of glucose in the human body. Diabetes, if not detected early, can cause a variety of issues in the human body, including heart disease, renal disease, eye damage, nerve damage, and so on. Machine learning is one of the most effective and precise methods for predicting diabetes in the human body. On a dataset constructed with PIMA samples, this system used several machine learning approaches. This study implements and compares Naive Bayes and Random Forest methods to detect diabetes, as well as highlight the issues encountered in each of these models. NB and RF methods can work on multiple input variables of PIMA dataset to derive prediction of diabetes. The system performance is measured with accuracy of outcome using different performance metrics.