Enhancing Diabetes Prediction Accuracy with AdvanSVM: A Machine Learning Approach Using the PIMA Dataset


Asha V , Manjunath Ramanna Lamani,Padmaja K , Tejashwini A Gondhale

 DOI10.5110/77. 1412               Page:   51-72         Vol: 19    Issue: 05   Year: 2024

creative commons, cc, character-785334.jpg   


Current research introduces the AdvanSVM, an improved Support Vector Machine designed to refine diabetes predictions using the PIMA Indian Diabetes dataset. The study explores various imputation methods to assess their impact on predictive precision, underscoring the significance of feature selection for fine-tuning machine learning models targeting diabetes forecasts. It also points out the importance of data balance and sophisticated model formulation, along with the need for broader datasets to propel forward strides in this field. Challenges pertaining to the specificity of the dataset and the extension of the results beyond its scope are acknowledged. The AdvanSVM classifier, featuring custom kernel functions and tailored parameter adjustments, addresses these issues while also managing the inherent imbalance found in medical data. The research targets the unique challenges of the PIMA dataset, such as missing information and anomalies, striving for improved, clinically applicable predictions of diabetes. A thorough evaluation of the model with relevant metrics confirms its potential for providing precise diabetes predictions.


Diabetes Prediction, Machine Learning, Support Vector Machine, AdvanSVM, PIMA Indian Diabetes Dataset

Received: 19 April 2024

Accepted: 10  May 2024

Published: 19 May 2024