A Decision-Making Model for Selecting Criterion in a Decision Tree Algorithm

Abstract

The decision tree algorithm is widely used for classification, which is a part of supervised learning. It classifies the node based on criterions for information gain which are termed as Entropy and Gini index. A comparative study of these techniques is required to determine the most feasible criterion for the specific dataset. As of now, machine learning and healthcare go hand in hand. Diabetes is a common disease that occurs when blood glucose levels exceed the normal range. It is expected to affectmillions of people, with half of the population remaining undiagnosed. The medical field generates a large amount of data, which is required to be analyzed further using machine learning. This paper focuses on machine learning, specifically the Supervised Learning’s algorithm decision tree, which is used for classification, to predict whether a patient is diabetic or not. When using a classification technique, the criterion plays an important role. Here, two popular criterions are discussed and compared for the accuracy of both. Ultimately, the prediction is based on a comparative study of both the Gini Index and the entropy to determine the best way to gain more accuracy. The algorithm is used on dataset, which is fed with different parameter values to get the result of the prediction.