We started out trying to make an application and model that would be able to approve credit card eligibility for customers however, we were not able to find any relevant data for that. After browsing on data.gov for a while, we found loan data and decided to change our project to do that instead.

It uses artificial neural networks to create a predictive model based on existing data and use the model for future data. The employee would put in the customer's information into the web page and click submit. The model would be called and process in the input variables and give back a "Yes your loan would be approved" or "No, it will not."

In MatLab, we used a non-linear multi-layer model with 5 hidden layers and got approximately 95% accuracy. We also used Sublime text to build a html page with css elements.

After building it in Matlab we attempted to recreate the same model in python to be able to integrate with Swift and xCode however we were told that django would be a better option. But by the time we found out about django it was a bit too late to implement it due to the time constraint.

We are really proud to have generated a model with 95% accuracy

We had never used xCode before so it was nice to learn more about it and be able to generate and test a dummy app and experiment with it.

Ideally, put together it would help employees provide customers with a quicker answer to their loan request without having to go through the whole process. We would also like to be able to give a percentage of approval rating in the future rather than a just yes/no.

Share this project:

Updates