EduScore

EduScore: Empowering Your Financial Future, One Score at a Time!

Comment

Inspiration

One of our team members shared this inspiration: "I first applied for a Discover student card the moment that I turned 18 to take advantage of a $20 'good grades' bonus. The credit card also included a feature that allowed me to view my credit score through Experian. However, it didn't explain how I could improve the score, or allow me to experiment with different situations to see how my credit score would respond." More generally, we have a desire to help college students take control of their economic situations. This is especially important right now, when job prospects, especially in tech, are not as rosy as they were in the past years. We see this app having the potential to really help students be fiscally responsible and lead them to make smart financial decisions going forward so that they can really succeed.

What it does

We have a web application where you input basic financial information and then receive a prediction for your credit score based on a proprietary machine learning model. It then gives you information about how you can improve your credit score.

It does this without risking your social security number or giving your data to third parties. It also does not affect your credit limit at all. In this way, it provides a safe way to learn and experiment.

Furthermore, it does this all securely with strong authentication protocols.

How we built it

We found a high-quality data set of labeled credit score data. We then pre-processed the data and trained a model using Scikit-Learn's implementation of the support vector machine (SVM) algorithm. We also did hyper-parameter tuning and dropped insignificant features. We then exported this model and uploaded it onto our backend webserver, which is hosted on PythonAnywhere. This backend webserver has an API that responds to a form put on the user-facing website.

The authentication is done through PropelAuth, which allows you to securely login using commonly used accounts such as Google and LinkedIn.

Challenges we ran into

We ran into significant challenges with putting our model onto a deployable webserver. It was easy to have the model run on a local Flask environment, but finding one that would work for free on the public internet was difficult. We ended up using Python Anywhere. We spent a bit of time debugging different file formats to store the model in. We realized that we had to export our model as a "pickle" file. (Pickles are always the answer to life's challenges.) We also dealt with CORB/CORS security issues. We fixed these by changing around our HTTP headers. Furthermore, the DNS configuration took a bit of work. We are grateful for the help of (Go)Daddy.

Accomplishments that we're proud of

We're really proud of the product we produced and how much we learned about back-end web development! We are proud that our credit score website could be of meaningful use in the hands of anyone trying to learn their way with the credit system and delighted with our team's collaboration. We're proud of how we worked together, especially given that we all came from different schools, years in college, and amounts of experience. We are especially proud of how it all came together after less than 24 hours!

What we learned

We learned about Flask and backend web development more generally. We also learned more about the HTTP protocol. Furthermore, we learned more about PropelAuth and configuring DNS settings. We also improved our soft-skills by working together, delegating and dividing up tasks, and coordinating work flows.

What's next for EduScore

We wish to provide more information to help people improve their credit score. Furthermore, we can integrate with other financial services to include more information in our model. Our aim will also be to refine the website to support higher service loads, so that more students and individuals desiring to use a credit score application can have the proper access respectively.

Sponsor Utilizations

We used a domain name from GoDaddy Registry called eduscore.courses. The challenge that this focuses on is the MLH - Best Domain Name from GoDaddy Registry, as we tried to include a relevant and fun name for our credit score website. We used PropelAuth to add a layer of security and setup an effective login system for any users interested in accessing the website. The challenge that this focuses on is the MLH - Best Use of PropelAuth, as we aimed to implement a proper setup for users to enter their credentials and be verified into the eduscore.courses either through direct login or social login.

Share this project:

Updates