Free Food Radar

Inspiration

We got the idea from geek story that an outstanding geek get free food every day by analyzing the event emails. It's far more a story because we searched but only got a free food info website without any information. So we decide to make it reality.

What it does

Free Food Radar is a free-food information collector and broadcaster. It collects all information of events from umich website, analyzes them and pick up the ones with free food. It shows all free-food event online with detailed time, location and source website. And it auto replies the request for a nearby free food.

How we built it

With a web crawler, event information are acquired and sent to Machine Learning analyzing engine. The analyzing engine give out the probability whether the event will provide free food, and send them to online server. Online server gathers all free food information into database and list them online. Google map API helped a lot in showing the location of events. Also, with twilio API, the server will recommend the nearest free food for users through message.

Challenges we ran into

The main problem was that we didn't have a data set to train the neural network. We spent nearly half a day on crawling data and labeling for them manually. But even we try hard to gather data, the data on umich website were still limited. The network were easy to be over-fitting.

Unfamiliar API is another difficulty for us. To make a wonderful project, we need many functions that we have never tried before. To apply google map and SMS service into our project, we spend much time on reading documents. Fortunately, it seems to be worthy.

Accomplishments that we're proud of

Firstly, we have trained a great neural network to predict probability of free food with limited data set. Of all 222 events in the test set, there are only two false-negative and there are no false-positive at all. The reliable engine is the solid foundation of the whole project.

Also, with the project, we provide user-friendly services. With our website, user can efficiently catch the free food information, or make long-term plan of taking part in those events. User can even get the free food info without accessing to the internet. Besides, this will be good propaganda for those events, and will lead to more free food, which is a win-win situation.

What we learned

Within the project, we all have fully polished our coding skill of solving specific problems. Instead of doing pieces work like in courses, this time we finished the whole solution for getting free food. The limited time impelled us moving forward and finally we made it.

What's next for Free Food Radar

Next we plan to apply it onto another university's website. It will bring the model more data to train, and bring the chance of free food for more students. Also, improving the SMS service with more functions and more personalized services is another possible aspect to work on.