Explainable Deep Ensemble Learning for Improved Credit Default Prediction
Year: 2024
Type: conference-paper
Abstract: Credit default prediction is a critical task in managing financial institutions. The need to enhance risk anticipation models’ overall robustness, and performance while ensuring these sophisticated models remain transparent and interpretable has grown. This paper proposes an approach for credit default prediction by leveraging several ensemble deep neural network models. Furthermore, it examines various Explainable artificial intelligence techniques to help us understand and interpret predictions made by the deep learning learning models. The results of our review show that the ANN-LSTM ensemble model performed better than the rest of the individual models with a precision of 0.84, recall of 0.95, and F1 score of 0.89.
Authors: Kagaba Dennis, Busingye Caroline, Barbara Nansamba, Daudi Jjingo, Ggaliwango Marvin
Institution: Makerere University
Language: English
Subfield: Accounting
Domain: Social Sciences
Open Access status: gold
OpenAlex ID: W4403882684