Exploration on Predicting Diabetes with Survey Responses

Abstract

Diabetes has become an increasingly ubiquitous problem, especially in the United States. However, it typically necessitates individuals to visit a medical center and undergo various measurements based on complicated metrics for a diabetes diagnosis or risk prediction. This paper attempts to explore the use of a 2015 survey to predict whether a person may develop diabetes or prediabetes using machine learning. This is a final project for the Applied Machine Learning course at Carnegie Mellon University. The results may offer inspiration and insights to healthcare providers.