Show simple item record

dc.contributor.advisorAlam, Md Ashraful
dc.contributor.authorAbrar, Tahmid
dc.contributor.authorTasnim, Samiha
dc.contributor.authorHossain, Md. Mehrab
dc.date.accessioned2019-10-29T10:13:07Z
dc.date.available2019-10-29T10:13:07Z
dc.date.copyright2019
dc.date.issued2019-09
dc.identifier.otherID 14301051
dc.identifier.otherID 18341027
dc.identifier.otherID 131010 43
dc.identifier.urihttp://hdl.handle.net/10361/12817
dc.descriptionThis thesis is submitted in partial fulfillment of the requirements for the degree of Bachelor of Science in Computer Science, 2019.en_US
dc.descriptionCataloged from PDF version of thesis.
dc.descriptionIncludes bibliographical references (pages 41-44).
dc.description.abstractChronic kidney disease (CKD) is a global prevalent ailment that causes lives in a predominant number. CKD is the 11th most deadly cause of global mortality with 1.2 million death each year and according to kidney Foundation of Bangladesh, around 40,000 CKD people experienced kidney failure annually as well as several thousand passed away in short stage of life because of CKD. Predictive analytics for healthcare using machine learning is a challenged task to help doctors decide the exact treatments for saving lives. Scientist researched collaboratively chronic kidney diseases, with the majority of their work on pure statistical models, generating numerous gaps in the development of machine-learning models. In this article we discussed the current methods and suggested improved technology based on the XGBoost (Extreme Gradient Boost), which combined significant characteristics of the F scores and evaluated four pre-processing scenarios. In addition, we provided machine training methods for anticipating chronic renal disease with clinical information. Four techniques of master teaching are explored including Support Vector Regressor (SVR), logistic Regressor (LR), AdaBoost, Gradient Boosting Tree and Decision Tree Regressor. The components are made from UCI dataset of chronic kidney disease and the results of these models are compared to determine the best regression model for the prediction. From this four preprocessing cases, replacing missing values with mean values of each column and choosing important features was most logical as it allows to train with more data without dropping. However, XGBoost gave the best outcomes in all four cases where it obtained 98% accuracy in case one where nulled valued are dropped, 98.75% testing accuracy for both case two and three where null values were replaced with minimum and maximum values of each column and it scores 100% accuracy in case four where null values are replaced with mean values. Thus, the system can be implemented v for early stage CKD prediction in a cost efficient way which will be helpful for under developed and developing countries.en_US
dc.description.statementofresponsibilityTahmid Abrar
dc.description.statementofresponsibilitySamiha Tasnim
dc.description.statementofresponsibilityMd. Mehrab Hossain
dc.format.extent44 pages
dc.language.isoenen_US
dc.publisherBrac Universityen_US
dc.rightsBrac University theses are protected by copyright. They may be viewed from this source for any purpose, but reproduction or distribution in any format is prohibited without written permission.
dc.subjectChorionic kidney diseaseen_US
dc.subjectXGBoosten_US
dc.subjectSupport Vector Regressor (SVR)en_US
dc.subjectLogistic Regressor (LR)en_US
dc.subjectAdaBoosten_US
dc.subjectGradient Boosting Tree and Decision Tree Regressoren_US
dc.subject.lcshMachine learning
dc.titleEarly detection of chronic kidney disease using machine learningen_US
dc.typeThesisen_US
dc.contributor.departmentDepartment of Computer Science and Engineering, Brac University
dc.description.degreeB. Computer Science


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record