Akademik Veri Yönetim Sistemi
Tezin Türü: Yüksek Lisans
Tezin Yürütüldüğü Kurum: Orta Doğu Teknik Üniversitesi, Enformatik Enstitüsü, Sağlık Bilişimi Anabilim Dalı, Türkiye
Tezin Onay Tarihi: 2014
Öğrenci: NEGİN BAGHERZADİ
Danışman: AYBAR CAN ACAR
Özet:The objective of this thesis is to develop and analyze the performances of a number of classifiers in prognosis classification based on a medical history data set. Generally, data mining uses algorithms originating in different disciplines such as artificial intelligence, statistics, optimization, database theory etc., to clarify available data. In this study, Support Vector Machine and Bayesian Network methods have been used. The data analyzed are clinical pathology records of patients with esophageal cancer who received an esophagectomy operation between 2003 and 2011. A large number of prognostic factors have been considered to classify the patients on prognosis. These factors found to be predictive were age, sex, dysphagia, odynophagia, lost weight, vomit, nausea, pathological N, pathological T, FEV1, tumor grade and tumor length. Classification trials for Support Vector Machine have shown %72.38 training accuracy with a generalization accuracy of %70.58, which was established by cross-validation. Support Vector Machine was used as the first method for data and SVM was found to achieve good accuracy on a data set of 119 patients when used in conjunction with PCA; SVM can be helpful for black-box analysis when data are irregularly distributed and produce accurate classifiers. Bayesian Network is the second method that is used in this study to solve missing data problem in the previous method. Bayesian Network can classify several different types of variables simultaneously. The quality of the results of the network that has been created for this study depends on the quality of the model. Performance of the model with Bayesian Network that is used in this study is 73.10% for 119 patients.