Remove implicit bias in risk-based mortgage loan models

Inspiration

Inspired by news that says banks are fined for $400M because of biasing interest rates against black and hispanic.

What it does

We explored the implicit biases in feature weighting, feature choices, and training data

How we built it

Challenges we ran into

Difficulty in obtaining open-source baseline models; and difficulty in obtaining data with racial labels.

Accomplishments that we're proud of

We are standing at the far-front of identifying the implicit biasing in models and proposed potential solutions.

What we learned

How to think the problems in the shoe of managers of the bank.

What's next for Remove implicit bias in risk-based mortgage models

Find baseline model and good training data!

Built With

• python