PPGCCM PÓS-GRADUAÇÃO EM CIÊNCIA DA COMPUTAÇÃO FUNDAÇÃO UNIVERSIDADE FEDERAL DO ABC Phone: 11 4996-8337 http://propg.ufabc.edu.br/ppgccm

Banca de QUALIFICAÇÃO: LEANDRO CARDINAL DOS SANTOS

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
DISCENTE : LEANDRO CARDINAL DOS SANTOS
DATA : 09/10/2020
HORA: 09:00
LOCAL: por participação remota
TÍTULO:

Predicting Credit Card Default using Machine Learning and Survival Analysis


PÁGINAS: 50
GRANDE ÁREA: Ciências Exatas e da Terra
ÁREA: Ciência da Computação
RESUMO:

The default of Brazilian consumers increased considerably in 2019. Consumers in this situation have difficulties obtaining credit,  affecting  directly entrepreneurship and preventing the generation of new jobs. Currently, the credit card industry is responsible for most customers' debt. The main objective of this work is to study computational methods that  predict consumer defaults. Default forecasting is modeled as a survival analysis problem, where the failure time is represented by the moment of non-payment. One of the differentials of this work is the adoption of a feature called the bill acquittance factor, which is calculated for each month of the experiment. This feature is used in the processing of the model which involves: (i) traditional statistical techniques; (ii) machine learning techniques using support vector machines. The evaluation of the model is a critical point of the study. Traditional performance metrics are not valid when the sample is censored. Thus, we will adopt the Concordance Index (C-Index) as a basis for comparison between the methods. The sample consists of data from a credit card company that operates in southeastern Brazil. The experiment was divided into two stages, where the first considered the discharge factor during the initial three months, while the second considered during the initial five months. A larger c-index was observed in the second stage, suggesting that this may be a relevant feature in preventing default.


MEMBROS DA BANCA:
Presidente - Interno ao Programa - 1773182 - SAUL DE CASTRO LEITE
Membro Titular - Examinador(a) Interno ao Programa - 1934625 - JESUS PASCUAL MENA CHALCO
Membro Titular - Examinador(a) Externo à Instituição - FABRÍZZIO CONDÉ DE OLIVEIRA - UFF
Membro Suplente - Examinador(a) Interno ao Programa - 3008017 - DENIS GUSTAVO FANTINATO
Notícia cadastrada em: 11/09/2020 22:01
SIGAA | UFABC - Núcleo de Tecnologia da Informação - ||||| | Copyright © 2006-2024 - UFRN - sigaa-2.ufabc.int.br.sigaa-2-prod