Research Article | Open Access
CLASSIFICATION OF DIABETES MELLITUS IN HUMAN WITH CUSTOMIZED K-NEAREST NEIGHBORS ALGORITHM IN COMPARISON WITH SUPPORT VECTOR MACHINE ALGORITHM
Bayinedi Nagarjuna, Mary Joy Kinol.A
Pages: 5547-5555
Abstract
Aim: The research work presents the classification of diabetes mellitus in humans with a Novel k-nearest neighbors algorithm in comparison with a support vector machine to improve the accuracy and precision.
Materials and Methods: Pima indian diabetes dataset has been collected from Kaggle, a machine learning repository for this study. The dataset contains 8 attributes that are considered as input attributes to the classifiers.
The Novel k-nearest neighbors algorithm and support vector machine are tested using 20 samples for each group.The significance value is <0.05. The G power is taken as 0.8. Results: K-nearest neighbors algorithm achieved better accuracy and precision of 94 % and 97 % compared to 92 % and 95 % of support vector
machine. Conclusion: In this research, it is found that the Novel k-nearest neighbors algorithm performed better accuracy and precision than the support vector machine algorithm in diabetes mellitus classification of the
dataset considered
Keywords
NovelK-Nearest Neighbors Algorithm, Support Vector Machine, Diabetes Mellitus, Blood Glucose Level, Machine Learning, Classification of Diabetes.