Prediction of Diabetes using Machine Learning

Exceptional progress in technology has led to significant production of data in healthcare. By employing various data analysis approaches it is possible to detect changes in the human body which will lead to early detection of diseases. Diabetes is one such disease that should be caught early in order to limit its advancement and to avoid other complications. In this work, machine learning algorithms are used to build a model to predict if a person suffers from type 2 diabetes or not. Logistic regression classification algorithm is applied on PIMA Indian Diabetic Dataset that consists of 8 parameters. The PIMA Indian Diabetic Dataset consists of 768 patient’s record out of which 268 patients have diabetes and other 500 do not have diabetes. Ensemble model is developed for classification purposes. Stacking technique is in cooperated here to use the predictions from the base model to build a new model to provide enhanced performance and accuracy than the base classifier alone. Other machine learning algorithms like Naive Bayes, Support Vector Machine Classifier and Decision Tree are used to build the model. Accuracy and execution time in each method is recorded in order to make better comparisons. The proposed method is implemented using python IDE and an accuracy of about 94% was achieved.