Credit Risk Model - Stardance

Ship #1 Changes requested

@knightfallShashy on Credit Risk Model · 25 days ago

Built a credit risk model that predicts the probability of default for loan borrowers using logistic regression. The most challenging part was understanding why logistic regression works for this problem instead of linear regression — the sigmoid curve keeps outputs between 0 and 1, which is essential for probabilities. I'm proud of building a fully functional expected loss function from scratch that a real bank risk team could use. To test it, run the script with the CSV in the same folder and call expected_loss_func() with any borrower's details to get their expected dollar loss.

2 devlogs
2h

Try project → See source code →

Open comments for this post

@knightfallShashy on Credit Risk Model · 25 days ago

17m 13s logged

I officially finished the Dist file. with everything compiled in it. Additionally, I’m done writing the README and pushed everything to Git.

Open comments for this post

@knightfallShashy on Credit Risk Model · 28 days ago

1h 22m 7s logged

Built a credit risk model for JPMorgan Chase Quantitative Research Task 3. Used logistic regression in Python (scikit-learn) to predict the probability of default (PD) for 10,000 loan borrowers. Split the data 80/20 for training and testing, then calculated expected loss per loan using the formula: PD × Loan Amount × 0.90, assuming a 10% recovery rate.