Page 614 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

Integrated statistical tests (e.g., demographic parity, disparate impact ratio) to ensure threshold
updates do not introduce or exacerbate bias, with regular reporting to regulators.

• REQ-08: Visualization Dashboard

Interactive dashboard (e.g., within Google Colab or web User Interface(UI)) displaying current
values of threshold levels, profit curves, feature importances, and MCMC trace plots for
stakeholder validation.

4 Sequence Diagram

5 References

[1] https:// www .kaggle .com/ c/ home -credit -default -risk/ data: Home Credit Default Risk

[2] https:// archive .ics .uci .edu/ dataset/ 144/ statlog+ german+ credit+ data: This dataset classifies
people described by a set of attributes as good or bad credit risks. Comes in two formats (one
all numeric). Also comes with a cost matrix

[3] M. Herasymovych and K. Märka, "Optimizing acceptance threshold in credit scoring using
reinforcement learning," Master’s thesis, Univ. of Tartu, Faculty of Social Sciences, School of
Economics and Business Administration, supervised by O. Lukason, 2018.

[4] M. Khashei and A. Mirahmadi, "A soft intelligent risk evaluation model for credit scoring
classification," *Int. J. Financ. Stud.*, vol. 3, pp. 411-422, 2015.

[5] J. L. Leevy, J. M. Johnson, J. Hancock, and N. Tran, "Threshold optimization and random
undersampling for imbalanced credit card data," J. Big Data, vol. 10, no. 58, pp. 1-13, 2023.
doi: 10.1186/s40537-023-00738-z.

[6] E. S. Kamimura, A. R. F. Pinto, and M. S. Nagano, "A recent review on optimisation methods
applied to credit scoring models," *Journal of Economics, Finance and Administrative Science*,
vol. 28, no. 56, pp. 352-371, 2023. doi: 10.1108/JEFAS-09-2021-0193.

[7] S. Kyeong and J. Shin, "Two-stage credit scoring using Bayesian approach," *J. Big Data*,
vol. 9, no. 106, pp. 1-18, 2022. doi: 10.1186/s40537-022-00665-5.

578