Journal of Computer Science
and Information Technology

Feature Selection and Parameter Optimization of Support Vector Machine (Svm) and Logistic Regression (Lr) Algorithms Using Particle Swarm Optimization (Pso) In Prediction of Diabetes
Aminu Usman Jibril, Khalid Haruna1 and Zhang Jiangsheng

Abstract
The rate at which people are suffering and dying as a result of diabetes is worrisome and devastating and ispredicted to rise to 11.3% by 2030 and to 12.2% by 2045 of the world populations are expected to have diabetes. Technological advancement has brought many improvements in the healthcare sector and other aspects of life. Diabetes is a chronic disease that stands among the top 10 deadliest diseases in the world for many years with no known cure and only early diagnosis and detection is the way forward for now which caught the attention of several Computer Scientists to use machine learning (ML) algorithms techniques in diabetes prediction. Algorithms are used during Coronavirus outbreak to create the vaccine formulation when the Chinese researchers release the genetic sequence of the virus on9th January, 2020. INOVIO researchers insert the genetic sequence of Coronavirus in their algorithm and came up with the experimental vaccine within 3 hours which led to the production and distribution of the vaccine within short period of time which indicates that the power of algorithms in today’s world cannot be overemphasized. In this research, we proposed the use of Particle Swarm Optimization (PSO)algorithm in optimizing Support Vector Machine (SVM) and Logistic Regression (LR) algorithms using PIMA Indian Diabetes dataset. The research contribute to the paradigm of knowledge by building and implementing the optimized models in diabetes prediction and got an accuracy of 98.67% in the optimized Support Vector Machine (PSO_SVM)and 97% accuracy in the optimized Logistic Regression (PSO_LR) respectively because to accurately diagnose the diabetes disease is the prime concern for doctors because is a matter of life and death which makes this model ideal for diabetes prediction.

Full Text: PDF     DOI: 10.15640/jcsit.v11n1a3