Intelligent Models for Diabetic Prediction Using Conventional Machine Learning Techniques and Ensemble Learning Algorithms

Abstract

The discovery of knowledge from medical database using machine learning approach is always beneficial as well as challenging task for diagnosis. Diabetes if left undiagnosed can affect many other organs (e.g., kidney and liver) of human body and this particular disease is very common in all ages young to adult. A large number of researches have been already taken place to predict diabetes using traditional machine learning algorithm such as artificial neural network, Naïve Bayes theorem, decision tree, etc. However, improvement of performance measures towards accuracy of identification of diabetes with a certain degree of confidence is a challenging task. Ensemble learning approach of classification of diabetes is one of such techniques in the parlour of machine learning classifier algorithms that provide a research gap for predicting the diabetes. This work presents classification algorithms for the prediction of diabetes based on two conventional machine learning classifiers (Naïve Bayes classifier model and decision tree) and four ensemble classifiers (Random Forest (RF), Bagging, AdaBoosting and Gradient Boosting). Performance measures of these algorithms have been carried out in terms of accuracy score. Dataset for training and testing the algorithms mentioned is retrieved from Pima Indian Database. On the basis of their comparative evaluation, most important feature with respect to identification of diabetic is extracted. This research underscores the significance of ensemble learning in diabetes prediction, comparing its efficiency with traditional classifiers. The study enhances accuracy assessment and identifies key features crucial for diabetes identification. These findings contribute valuable insights, paving the way for advancements in machine learning applications for healthcare diagnostics.