Inspiration

Our inspiration was creating a prediction software for a financial business company that could aid profit margins and correctly predict a human's behavior. Each of us had some sort of interest in the financial sector, so this project came together organically.

What it does

The Loan Advisor system is a python program that models a loanee's probability to pay the loan that they took out 6 months earlier.

How we built it

We used Google Colab and Jupyter Notebook for a majority of the project, sharing code through GitHub commits. We built this by creating a Random Forest model and assigning data to correctly predict loan payments.

Challenges we ran into

One great challenge we encountered was the inability to find a dataset that had FICO credit scores included with the. Credit scores would have been a great, easy way to predict if a loanee would pay back their loan after a specified time. To overcome this barrier, we found a dataset that has many of the same factors used to calculate a credit score, so we were able to create a consumer score in order to more accurately produce the result we originally wanted.

Accomplishments that we're proud of

One thing that we're incredibly proud about is the production of efficient and professional code, along with the idea that we persevered through many challenges in a 24 hour span. We originally started out as a group of 5, but 2 group members decided to give up half way through. This presented the challenge of trying to finish a project for 5 people with only 3 remaining members, which I believe makes the accomplishment of executable code even greater.

What we learned

Through a process of a combination of different fields; Machine Learning, AI modeling, and domain knowledge of loans/credit scores, I believe each of us have learned an incredible amount of knowledge. We've expanded our knowledge on Random Forest algorithms, financial domain facts, probability predictions, and more.

What's next for Loan Advisor Prediction Model

I think we're incredibly happy with the end result of the Loan Advisor system, but potentially in the future we would love to make this application more accessible. This would include creating a comprehensive GUI so people could input their own data to see how a prediction software runs against their own case.

Built With

Share this project:

Updates