A carregar...
Publicação
Application of Predictive Models for Bank Loans: Estimating a Reference Price to Improve the Loan’s Negotiated Amount
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 1.7 MB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
The banking sector has been significantly impacted by rapid technological advancements, prompting financial institutions to explore ways to adapt to these changes. This project resulted from an internship at Caixa Geral de Depósitos, where the primary objective was to assist the commercial chain in negotiating a specific type of loan with company clients. This was achieved by developing a predictive model that estimates a reference price for the loan’s negotiated amount based on the historical operations of similar clients. The price must address the client's risk and differ based on the type of guarantee provided. After analyzing the data, it was found that two separate models would need to be developed due to the presence of a group of products with business limitations. The project explored various algorithms, including Extreme Gradient Boosting, Random Forest, Gradient Boosting, CatBoost, and Support Vector Regression. Extreme Gradient Boosting was chosen as the optimal algorithm for both product groups, achieving an R2 of 0.844, an MAE of 18.26, and an RMSE of 33.78 in the normal credit product group, and an R2 of 0.68, an MAE of 18.49, and an RMSE of 33.43 in the group of products with business restrictions.
Descrição
Internship Report presented as the partial requirement for obtaining a Master's degree in Data Science and Advanced Analytics, specialization in Data Science
Palavras-chave
Bank Loan Client Risk Estimating Reference Price Extreme Gradient Boosting Machine Learning SDG 8 - Decent work and economic growth SDG 9 - Industry, innovation and infrastructure