Diabetes Diseases Prediction Using Supervised Machine Learning and Neighbourhood Components Analysis
Othmane Daanouni, Bouchaib Cherradi, Amal Tmiri
Abstract
Diabetes mellitus (DM) is a chronic disease, which can affect the entire body system. Early Diagnosis of patient's diabetics can help improve their health quality or reducing the risk factors. The main objective of this study is to evaluate the performance of some Machine Learning algorithms, used to predict diabetes diseases, for this purpose we apply and evaluate four Machine Learning algorithms (Decision Tree, K-Nearest Neighbours, Artificial Neural Network and Deep Neural Network) to predict diabetes mellitus. These techniques have been trained and tested on Pima Indian dataset. The performances of the experimented algorithms have been evaluated after removing noisy data and using features selection with Neighbourhood components Analysis in order to reduce the number of features and mitigate the complexity of dimensionality in favour of speeds up the learning process, enhances data understanding. Different similarity metrics used to compare model performance like Accuracy, Sensitivity, and Specificity.