A carregar...
Publicação
Credit Cards Application Scoring Model: Leveraging Machine Learning for Credit Card Applications
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 2.82 MB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
This thesis investigates the development and evaluation of credit scoring models using Logistic
Regression, XGBoost, and Random Forest to predict creditworthiness for credit card
applicants. The project aims to strike a balance between predictive performance, model
interpretability, and business impact, particularly in the context of highly imbalanced datasets,
where the number of good applicants vastly outweighs the number of bad ones. A structured
pipeline was built for each algorithm, including tailored preprocessing strategies and class
imbalance handling. Logistic regression used binned Weight of Evidence transformation and
multi-step feature selection, while ensemble models were trained on raw data with flexible
encodings, capturing non-linear patterns. To evaluate model performance, both conventional
classification metrics and a profit-orientated objective function were used, incorporating
realistic assumptions about profits from good clients and losses from defaults. Models were
validated in holdout and out-of-time samples under a unified evaluation framework. The
model outputs were translated into scores using a common scaling convention, allowing direct
comparison of score ranges and risk profiles between techniques. Logistic regression
supported the creation of a transparent scorecard, while explainability in tree-based models
was addressed using SHAP values. The results indicate that XGBoost slightly outperforms
other models in balancing predictive accuracy and profitability, while logistic regression excels
in transparency for scorecard use.
Descrição
Dissertation presented as the partial requirement for obtaining a Master's degree in Data Science and Advanced Analytics, specialization in Business Analytics
Palavras-chave
Credit Scoring Scoring Models Credit Risk Modelling Machine Learning Algorithms Credit Card Applications SDG 8 - Decent work and economic growth SDG 9 - Industry, innovation and infrastructure SDG 17 - Partnerships for the goals