Klasifikacija kreditne sposobnosti modelom logističke regresije: problem nebalansiranih kategorija

master's thesis

• Title (english): Classification of credit score by logistic regression model: the problem of unbalanced categories
• Author: Dora Akmačić
• Mentor: Snježana Lubura Strunjak (mentor)
• Committee member: Snježana Lubura Strunjak (predsjednik povjerenstva)
• Committee member: Petra Posedel Šimović (član povjerenstva)
• Committee member: Nikola Sandrić (član povjerenstva)
• Committee member: Igor Ciganović (član povjerenstva)
• Granter: University of Zagreb; Faculty of Science; (Department of Mathematics); Zagreb
• Defense date and country: 2024-07-17, Croatia
• Scientific / art field, discipline and subdiscipline: NATURAL SCIENCES; Mathematics

Abstract

Cilj ovog rada je klasificirati kreditnu sposobnost modelom logističke regresije. U prvom poglavlju obrađeni su najvažniji pojmovi potrebni za razumijevanje logističke regresije. Prvo je objašnjeno kako se može definirati logistička regresija pomoću generaliziranih linearnih modela i koja je uloga funkcije veze logit. Zatim je dano objašnjenje kako se metodom maksimalne vjerodostojnosti procjenjuju parametri modela. Takoder, dani su najvažniji rezultati i testovi koji se koriste prilikom prilagodbe modela podacima. Drugo poglavlje bazirano je na primjeni logističke regresije za klasifikaciju klijenata koji su kreditno sposobni i onih koji nisu. Nakon analize i čišćenja podataka u deskriptivnoj statistici, testirana je značajnost nezavisnih varijabli modela. Nakon odabira modela, dane su metrike kojima se ispitala točnost predikcija modela. Treće poglavlje uzima u obzir da podaci imaju problem nebalansiranih kategorija. Iz tog razloga uveden je algoritam SMOTE kao pomoćni alat koji smanjuje razliku izmedu broja opažanja dviju klasa. Na kraju se usporeduju rezultati sa i bez SMOTE-a.

Abstract (english)

The aim of this thesis is credit score classification using a logistic regression model. In the first chapter, the key concepts necessary for understanding logistic regression are covered. First, it explains how logistic regression can be defined using generalized linear models and the role of the logit link function. Then, an explanation is given of how the parameters of the model are estimated using the maximum likelihood method. Additionally, the most important results and tests used for the goodness of fit of the model to the data are presented. The second chapter focuses on applying logistic regression for credit score classification. After analyzing and cleaning the data in descriptive statistics, the significance of the independent variables in the model is tested. Following the model selection, metrics are provided to evaluate the performance of the model’s predictions. The third chapter addresses the issue of imbalanced data. For this reason, the SMOTE algorithm is introduced as an auxiliary tool to reduce the difference between the number of observations of the two classes. Finally, the results with and without the use of SMOTE are compared.

• Keywords: generalizirani linearni modeli; funkcija veze logit; algoritam SMOTE
• Language: croatian
• URN:NBN: urn:nbn:hr:217:667860
• Study programme: Title: Financial and Business Mathematics; Study programme type: university; Study level: graduate; Academic / professional title: sveučilišni magistar matematike (sveučilišni magistar matematike)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2024-07-24 10:21:58