Design of a Data Analytics Model to Predict Diabetes using Machine Learning Techniques

Diabetes Mellitus is a chronic disease that is characterized by high blood sugar level for a prolonged period of time. Diabetes is due to either insufficient secretion of insulin by the pancreas or the body cells not properly responding to the insulin that is produced. Improper management of the disease will often lead to severe health complications. Prediction of such disease can be automated using data mining techniques. The proposed data analytics system will use a hybrid classifier model that uses two different data mining classification techniques to classify the disease. The prediction of individual instance of data is done by a machine learning framework that uses Artificial Neural Networks. A Multilayer Perceptron Neural Network is created and it is trained using Back Propagation Algorithm that will classify the patients tested positive as 1 and patients tested negative as 0. This trained neural network gave a total mean square error of 0.1213 on test. The system thus produced a recognition rate of 80%.