Comparative Analysis of Machine Learning Algorithms for Diabetes Prediction

• Author(s): Omoshola Ogunduboye
• Paper ID: 1706978
• Page: 527-536
• Published Date: 23-01-2025
• Published In: Iconic Research And Engineering Journals
• Publisher: IRE Journals
• e-ISSN: 2456-8880
• Volume/Issue: Volume 8 Issue 7 January-2025

Abstract

Diabetes is regarded as one of the most chronic metabolic diseases if left unchecked. Diabetes is a worldwide chronic health issue. Today, approximately 400 million people are living with diabetes. A large percentage of people who are living with diabetes are unaware of their condition until it becomes chronic. Diabetes, also known as Diabetes Mellitus, is an increasingly prevalent chronic disease which affects the body’s ability to metabolize glucose. With the growing rate of diabetes cases, it has become important to take a deeper look into solutions and ways to better handle the situation. This paper presents a predictive approach to diabetes, through diabetes prediction using machine learning, a process that will allow for better treatment and preventive healthcare. Machine learning in diabetes prediction is important because there is a vast pool of available data on diabetes both through research and years of clinical studies. This data can be processed and fed into machine learning models to highlight meaningful relationships and patterns within patients’ data. However, this has been hampered by the difficult task of choosing the best machine learning algorithm. A challenge that can be solved by carrying out a comparative study using different evaluation metrics to ascertain which algorithm produces the most optimal results. This paper represents the result and analysis regarding detecting a person’s diabetic state from various machine learning models based on key attributes such as age, gender, glucose level and insulin level. The model proposed was achieved by collating diabetes data from Kaggle and prepossessed to remove abnormalities and irrelevant attributes after which it was divided into test and training data. The machine learning algorithms chosen for this study were SVM, logistic regression, decision tree, random forest classifier and K-Neighbors classifier. The best performing model was random forest with an accuracy of 95%. This paper contributes to the diagnosis and prediction of diabetes through the application of machine learning in predicting patients who are likely to live with diabetes.

Keywords

Diabetes, Decision Tree, Dataset, Attributes, Machine Learning, SVM, K-Neighbors, Random Forest Algorithms.

Citations

IRE Journals:
Omoshola Ogunduboye "Comparative Analysis of Machine Learning Algorithms for Diabetes Prediction" Iconic Research And Engineering Journals Volume 8 Issue 7 2025 Page 527-536

IEEE:
Omoshola Ogunduboye "Comparative Analysis of Machine Learning Algorithms for Diabetes Prediction" Iconic Research And Engineering Journals, 8(7)