Prediction of Type 2 Diabetes using Machine Learning Classification Methods

Over 30 million people in India are suffering from diabetes and many others are under the risk. Thus, early diagnosis and treatment is required to prevent diabetes and its associated health problems. This study aims to assess the risk of diabetes among individuals based on their lifestyle and family background. The risk of Type 2 diabetes was predicted using different machine learning algorithms as these algorithms are highly accurate which is very much required in the health profession. Once the model will be trained with good accuracy, then individuals can self-assess the risk of diabetes. In order to conduct the experiment, 952 instances have been collected through an online and offline questionnaire including 18 questions related to health, lifestyle and family background. The same algorithms were also applied to the Pima Indian Diabetes database. The performance of Random Forest Classifier is found to be most accurate for both datasets.