Big Belly Monitor

Inspiration

After spending half the hackathon working to understand some of the coarser points of machine learning, I wanted to tackle a meaningful project that involved the technology. Perusing through the data sets at Analyze Boston, I decided to start developing a utility that would help keep my city clean and beautiful.

What it does

Using historical data, this application can potentially predict the usage of a BigBelly trash compactor at an arbitrary location and time.

How I built it

I pretty much stuck to the detailed documentation and easy to follow tutorials provided by the TensorFlow developers.

Challenges I ran into

I'd never touched machine learning before yesterday; it was an extremely steep learning curve.

Accomplishments that I'm proud of

I actually got output! I find it incredible that what used to be (and still is) the realm of hardcore computer science research is also so easily accessible that I could generate output over the course of one night.

What I learned

Machine learning is HARD! I am building a tenuous grasp on the theory behind the operation; I'm also starting to wrap my head around some of the nebulous features in TensorFlow.

What's next for Big Belly Monitor

Due to my unfamiliarity with TensorFlow, I had to strip some information out of the input. Hopefully, I can add that back and get more meaningful output.

Built With

• analyze-boston
• python
• tensorflow