COBA (Capital One Budgeting Application)

Inspiration

The need for a smarter budgeting app as a college student and future young professional.

What it does

The tool categorizes bank account holders into buckets according to their personal profile (College Student, Young Professional, Mid-Life, Retirement). The tool analyzes the spending habits of the users in each spending profile and selects the “optimal budgeters”. It then uses these accounts’ spending habits as a benchmark to suggest new budget category ranges to users who fit these profiles.

How we built it

We created a python jupyter notebook to follow the flow of the data analysis. Since there is no publicly available bank account data (for privacy reasons), we created 10000 randomized dummy bank accounts for testing.

We used Python’s Pandas data analysis library to isolate and analyze the data. We used GitHub for collaboration and code sharing. We used some of Microsoft Excel’s data tools as well.

Challenges we ran into

We initially wanted to use Capital One’s recommended Nessie API. However, it proved to be limited. We would have liked to use Google’s Cloud Machine Learning Engine to drive the data analysis and prediction, but it proved unrealistic given our time and knowledge constraints.

Since we did not use ML for this project as initially intended, we hard-coded some of the identifiers of good budgeting profiles. This is something that we did only for the purpose of the demonstration and would not be the method going forward.

Accomplishments that we're proud of

Attending our first hackathon and having a real problem-solving experience. We initially were not planning to compete, so to actually end up building something that we like is something we are proud of. We are proud of the idea and the thought process behind its formation and potential validity for the market. We were able to thoroughly plan out a plan of action on the functioning and deployment of an app.

What we learned

This was our first hackathon!

We experienced many roadblocks such as multiple partners dropping and switching teams. We ourselves were not sure about how committed our spirit was.

We went to Capital One’s workshop as well as a venture capitalism workshop and were inspired to build. We came up with a basic principle of: “Profile and Suggest” and this is what led us to come up with the project, where we profile a user and offer suggestions to help them budget. We grew as roommates in this activity, experienced a night without sleep, struggled to compromise but found ways to in resolution, spoke our minds, and for once did not mindlessly consume but rather produced technology.

What's next for COBA

We want to build a solid front-end user interface and have actual data on users to run analytics. We want to use machine learning to classify an optimal budgeters and provide smarter suggestions. We want to improve relationships with customers and increase privacy and user choice in the process.

Built With

Share this project:
×

Updates