Data visualization and pre-processing techniques based Diabetes Prediction System

Over time, having too much glucose in blood can cause health issues. Diabetes occurs when your blood glucose or sugar is too high. One in six people with diabetes in the world is from India. With the development of standards of living, diabetes is gradually increasing in people. Therefore, a convenient way to predict diabetics is essential so that necessary protocols can be taken beforehand. With this intention, we developed a web-based diabetes prediction application which helps its users diagnose the possibility of this disease by entering valid data. In this paper, a diabetes prediction system is implemented for predicting diabetes which comprises of some external features for diabetes alongside regular features like Insulin, Glucose, BMI, Age, etc. This application is designed with the help of HTML and CSS. The prediction model, which produces highly accurate results, applied and compared various algorithms like K-nearest neighbors (KNN), Decision Tree, Logistic Regression (LR), Random Forest (RF), Support Vector Classifier (SVC) and Naïve Bayes to determine Diabetes. The database system for our web application is constructed using My SQL with the vision of creating a system where a user can log on to the system for personal use. The entire application is deployed using XAMPP. The system utilizes Flask framework in Python to launch the web application. To achieve improved accuracy on the datasets, Data visualization and pre-processing techniques were taken to use. One of the data attributes to be inputted requires user's BMI record and to calculate that the application also contains an in-build BMI calculator.