Modelo de sistema de recomendação para recuperação de crédito com foco na equipe de cobrança
With the increase in indebtedness year after year, Brazil is not experiencing a good economic moment, not even for political, economic, health or other reasons. Despite this, the Brazilian people always seek to solve their financial problems. However, when debt payment is not possible, credit protec...
| Autor principal: | Cobiank, Douglas Teixeira |
|---|---|
| Formato: | Trabalho de Conclusão de Curso (Graduação) |
| Idioma: | Português |
| Publicado em: |
Universidade Tecnológica Federal do Paraná
2023
|
| Assuntos: | |
| Acesso em linha: |
http://repositorio.utfpr.edu.br/jspui/handle/1/30930 |
| Tags: |
Adicionar Tag
Sem tags, seja o primeiro a adicionar uma tag!
|
| Resumo: |
With the increase in indebtedness year after year, Brazil is not experiencing a good economic moment, not even for political, economic, health or other reasons. Despite this, the Brazilian people always seek to solve their financial problems. However, when debt payment is not possible, credit protection services become part of the process and debts are protested in court. Taking into account that the advancement of information technology and data intelligence enabled the development of tools that help people in their daily lives, this work proposes the construction of a model of recommendation of actions based on multi-label classification to facilitate the process. of debt collection. This model advocates actions that operators can carry out at the time of collection, which can result in greater assertiveness in contacting debtors, expanding the possibility of paying off debts. In this work, two models were created, with more labels and fewer labels, to perform the multi-label classification, the first, with fewer labels, being statistically better when compared to the second. The accuracy in the first model was on average 71%, which indicates that, if the collection team takes the recommended actions, there is a 71% probability that the debt will be paid off. In this way, the recommendation model based on multi-label classification achieved better results than the normal debt recovery rate, 56%, which demonstrates that it can be an alternative for collection teams at the time of credit recovery. |
|---|