A novel method for diabetes classification and prediction with Pycaret

Abstract

The incredible advances in biotechnology and public healthcare infrastructures have resulted in a massive output of vital and sensitive healthcare data. Many fascinating trends are discovered using intelligent data analysis approaches for the early identification and prevention of numerous severe illnesses. Diabetes mellitus is a highly hazardous condition since it leads to other deadly diseases such as heart, kidney, and nerve damage. In this research study, a low code Pycaret machine learning technique is used for diabetes categorization, detection, and prediction. On applying Pycaret various classifiers having different accuracies are produced and shown in the result section. After hyper tuning of various classifiers, it is found that the gradient boosting classifier is best further tuned and an accuracy of about 90% is achieved which is the highest among all existing ML classifiers.