Daemon Dash Hackathon 2015 - SmartSpenders

We took on Capital One's challenge at the Daemon Dash Hackathon to use their API, Nessie, in the best way we could.

We decided to try to create an way for parents to monitor the way their children spend money. Not only would this track all purchase prices, types of items bought, and merchant names, but it will also keep track of cummulative data. Using this, we can predict how a student will spend for the rest of the month, and can let a parent know whether or not his or her child will go over the monthly allowance provided.

This is not just linear regression, however. If we did it normally, impulse buys such as an $800 TV would ruin the slope of the line. Instead we check if purchases are outside 1.5 * std dev from the mean. If they are, the value of that purchase is included on the y axis, as it would control the y intercept.

This project was taken on by four freshman, two of whom are not even computer science majors. Three of the four members did not know python going into the project, and only one of the members knew web design. We originally planned to not just create the parent's interface, but also one for the child so he or she can track his or her own spending, but that proved to be too much for 25 hours with no sleep.

We used python for the backend/data analysis and the library "plotly" to create the graph. Nessie was used to come up with all of the data and purchases, and we could store all of the data on a server generated by Capital One. For frontend, basic CSS and HTML was used (CSS template: getskeleton.com). We had to use flask to integrate the python and HTML, and this proved to be one of the most difficult parts of the library. To send emails to the parents, an SMT server was established, which sends automatic emails as a child spends money.

I apologize for the sporading nature of the python files and everything else in this repo; such is the nature of a hackathon ;)