Diabetes Detection using Neighbourhood Component Analysis with SVM and Random Forest Classifier

Abstract

This paper describes the design and implementation of a computational model capable of classifying individuals with suspected Diabetes Mellitus. The model was developed using machine learning techniques and uses for training a database provided by a network of clinical analysis laboratories.Also, Neighbourhood Component Analysis (NCA) is utilized for selection of features from the PIMA Indian diabetes dataset. Classification of selected features is done by Random Forest Classifier and Support Vector Machine (SVM) algorithms. Performance evaluation of above-mentioned methods is carried out on the basis of evaluation parameters; Precision, Accuracy and Sensitivity.