596. Data-Driven Decision-Making in Credit Risk Assessment: A Novel Deep Neural Network Model
Invited abstract in session TD-28: Risk and Financial Decision Making, stream Decision Support Systems.
Tuesday, 14:30-16:00Room: Maurice Keyworth 1.03
Authors (first author is the speaker)
| 1. | Reza Ghasemy Yaghin
|
| Industrial Engineering and Management Systems, Amirkabir University of Technology | |
| 2. | Nima Mahmoodian
|
| Industrial Engineering, Amirkabir university of technology |
Abstract
This paper presents a novel six-layer deep neural network (DNN) framework designed to enhance credit risk assessment within a decision support system. Leveraging the Lending Club dataset, the proposed method integrates an extensive preprocessing pipeline—including feature elimination, categorical encoding, scaling, and random under-sampling—to effectively manage missing data and class imbalance. Four distinct DNN architectures were developed, each employing ReLU activation functions in the hidden layers and a sigmoid function in the output layer, along with elastic net regularization and a 20% dropout rate to mitigate overfitting. Comparative experiments reveal that one of the models, achieves superior performance, particularly in recall and F1-score, when benchmarked against both other DNN configurations and a contemporary stacking approach. Embedding the proposed framework into a decision support system enables financial institutions to harness data-driven insights for informed credit decisions and improved risk management. These results not only highlight significant improvements in prediction accuracy but also underscore the system’s potential to streamline credit scoring processes, ultimately providing a robust tool for strategic decision-making in financial operations.
Keywords
- Decision Support Systems
- Analytics and Data Science
- Financial Modelling
Status: accepted
Back to the list of papers