MoneyMore was written during the Fall 2014 PennApps hackathon to analyze a Venmo user's transaction data. I worked with two friends; one had just transfered to Computer Science and we wanted him to get the hackathon experience, but we didn't come into the weekend with any significant ideas. There were Venmo employees promoting their open API, so we set on trying to get some useful information out of it.
We ended up throwing together three ideas: identifying the person a user paid or charged the most, creating a graph of a user's Venmo social network, and a prediction for what a user's next charge would be.
The first, which we dubbed the "Venmo Lover" was eerily accurate. During demos, it would always guess a significant other if there was one to be had.
The network graph worked out well too, mirroring true social networks better than a friend graph on Facebook might -- if you are exchanging money with someone, chances are you are close! If there was a house with a big group of people living together, particularly fraternities or sororities, it would show as a large clique in the graph. When a Venmo employee came by during the demo, they found the company's graph had the two founders right in the center.
The predictive charge was a bit goofier. It didn't do anything spectacular under the hood, just took recent charges and randomly selected words for the message and a recipient. It was mostly just for laughs.
The project was originally called Venmore (which already didn't make much sense), but we got a very kind cease and desist email from Venmo shortly after the event, so we changed it to MoneyMore.