Please use this identifier to cite or link to this item: http://hdl.handle.net/10266/5202
Title: Predicting Diabetes Mellitus Using Machine Learning Based Ensemble Model
Authors: Mamta
Supervisor: Bhatia, Tarunpreet
Keywords: Machine Learning;Classification;Ensemble Model;Diabetes Prediction
Issue Date: 9-Aug-2018
Abstract: Diabetes is the fast growing disease among the people even among the youngsters. Diabetes is caused by the increase level of the sugar in the blood. Diabetes is a serious health problem that needs special attention and public health interventions in the 21st century. Diabetes Mellitus is defined as a chronic condition, a disorder of metabolism characterized by increased concentration of blood sugar levels caused by either insufficient secretion of insulin or resistance to insulin action or a combination of both. Diabetes is also the creator of another kind of disease mainly, the eyes, kidneys, blood vessels, nerves and the heart. Machine learning techniques are used in medical predictions. Machine learning allows building models to quickly analyze data and deliver results, leveraging both historical and real-time data. With machine learning, healthcare service providers can make better decisions on patient’s diagnoses and treatment options, which leads to the overall improvement of healthcare services. In this work, we have applied different machine learning models such as Decision Tree, Naive Bayes, Support Vector Machine, Random Forest, K-Nearest Neighbors, Adaboost, Linear Model and Neural Network. From these models, performance are evaluated on the basis of sensitivity, specificity, precision, recall, and accuracy. After that top five models are selected to perform ensembling. A Boosting based ensemble model is predicting to improve the accuracy of the dataset. The proposed ensemble model gives better results as compared to the single model in terms of accuracy. The accuracy is improved up to 84.82%. 10-fold cross validation is applied to improve the robustness of the data.
URI: http://hdl.handle.net/10266/5202
Appears in Collections:Masters Theses@CSED

Files in This Item:
File Description SizeFormat 
5202.pdf1.78 MBAdobe PDFThumbnail
View/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.