Akademik Veri Yönetim Sistemi
Tezin Türü: Yüksek Lisans
Tezin Yürütüldüğü Kurum: Gazi Üniversitesi, Fen Bilimleri Enstitüsü, Türkiye
Tezin Onay Tarihi: 2020
Tezin Dili: Türkçe
Öğrenci: Hamdi OKUR
Danışman: AYDIN ÇETİN
Özet:Recently, peer-to-peer credit systems have become widespread and emerge as an alternative to the traditional banking system. In order to give the right amount of credit to the selected customer quickly, the credit risks should be determined according to the customer characteristics and the amount of loan suitable for each customer group should be calculated according to the risk value. Lending Club is the most common peer-to-peer financial system. In this study, the factors affecting credit risk were investigated by testing the regression and binary classification methods in order to calculate credit risk with machine learning methods. In the thesis, regression and binary classification methods are trained and tested with the Lending Club data set. According to the test results, the most successful result among the regression methods was the decision tree regression algorithm, while in the binary classification type, the two-class decision tree algorithm gave the most successful results. In the general data analysis, the factors affecting credit risk were determined as income group, interest rate, region and credit use type according to priority level.