Makine öğrenmesi ve öznitelik seçim yöntemleri ile kardiyovasküler hastalıkların tespiti
No Thumbnail Available
Date
2023
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
İstanbul Ticaret Üniversitesi
Access Rights
info:eu-repo/semantics/openAccess
Abstract
Hastalıkların gün geçtikçe arttığı ve çeşitlendiği günümüz dünyasında belli başlı hastalıklar can kayıplarına neden olmaktadır. Bunların başında ise kardiyovasküler hastalıklar gelmektedir. Bu çalışmada kalp hastalığı tahminine yönelik UCI Machine Learning Repository veri tabanlarından Statlog veri seti incelenmiştir. Pearson Korelasyon analizi, Ki-Kare Testi ve RefiefF Algoritması kullanılarak öz nitelik seçim işlemi uygulanmış, seçilen öz nitelikler 7 ayrı senaryo üzerinde örneklenmiştir. Her bir senaryo örneklemi 6 ayrı yöntem kombinasyonu ile analiz edilmiş, toplam 2520 model sonucu incelenmiştir. En yüksek doğruluk olan %95 ile hiper parametre optimizasyonu ile İş Hattı mimarisi kombinasyonu ile Rassal Orman Algoritması ve Gradient Boosting algoritmaları ile ulaşılmıştır.
In today's world, where diseases are increasing and diversifying day by day, certain diseases cause loss of life. At the beginning of these, cardiovascular diseases come first. In this study, the Statlog dataset from the UCI Machine Learning Repository databases for heart disease prediction was examined. The self-attribute selection process was applied by using Pearson Correlation analysis, Chi-Square Test and RefiefF Algorithm, and the selected self-attributes were sampled on 7 separate scenarios. Each scenario sample was analyzed with a combination of 6 separate methods, and a total of 2520 model results were examined. The highest accuracy, 95%, was achieved with the combination of hyper parameter optimization and Pipeline architecture with the Random Forest Algorithm and Gradient Boosting algorithms.
In today's world, where diseases are increasing and diversifying day by day, certain diseases cause loss of life. At the beginning of these, cardiovascular diseases come first. In this study, the Statlog dataset from the UCI Machine Learning Repository databases for heart disease prediction was examined. The self-attribute selection process was applied by using Pearson Correlation analysis, Chi-Square Test and RefiefF Algorithm, and the selected self-attributes were sampled on 7 separate scenarios. Each scenario sample was analyzed with a combination of 6 separate methods, and a total of 2520 model results were examined. The highest accuracy, 95%, was achieved with the combination of hyper parameter optimization and Pipeline architecture with the Random Forest Algorithm and Gradient Boosting algorithms.
Description
Fen Bilimleri Enstitüsü, Bilgisayar Mühendisliği Ana Bilim Dalı, Bilgisayar Mühendisliği Bilim Dalı
Keywords
Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve Kontrol, Computer Engineering and Computer Science and Control
