AI Data Usage In The Justice System

Inspiration

What inspired was the racial injustice done to minorities and African Americans by the police.

What it does

It predicts whether or not a person is eligible to be released based on the prisoner's COMPAS scores. COMPAS is introduced and used in many jurisdictions around the U.S. to predict if a convicted criminal is likely to re-offend. COMPAS algorithms assign a score to each defendant ranging from 1 to 10 with ten being the highest risk

How we built it

We built on jupyter notebooks pooling our ideas and coming up with the best solutions. We trained many models like Logistic Regression, Linear SVC, Random Forests, Random Forests with grid search. Finally we decided to go with Random Search with grid search as it had the best f1 score (0.67)

Challenges we ran into

De biasing the model was the main challenge we had.

Accomplishments that we're proud of

We are proud that we have come up with a model in such a short time.

What we learned

We have learned to train models and make predictions based on real data. We also learned many different models and their different use cases.

What's next for AI Data Usage In The Justice System

We think implementing VAEs could de biase our model more.

Built With

• python
• sklearn