leafZer0

CO2 emissions over a set period as well as future predictions based off of the data.

Inspiration

What drove us to create a sustainability projects was the current state of Rochester. This once temperate climate has become highly volatile due to the increasing output of greenhouse gasses and has become apparent with this winter. As we all know, Rochester can change its mood in just a couple hours.

What it does

leafZer0 is a program that incorporates aspects of machine learning and artificial intelligence with a viable software product that encourages users to gain perspective on their contributions to greenhouse gas production. It takes in data from a user and outputs informative visual representations of the data alongside predictions concerning the trends followed by the data.

How we built it

This program was built with python. For the website, Django framework was used to develop the backend of the website while bootstrap was used for UI design. From the ML/AI side, the TensorFlow library was used to train and develop a model to perform analysis on. From this, linear regression and a long short-term memory model was used to implement a viable software solution.

Challenges we ran into

Developing a pure python application on paper sounds like should be fairly consistent and simple to work with. But both assumptions where incorrect. One of the most challenging problems we encountered was the inability to use javascript-esque web design for our app as Django lacks the front-end support for this. On the AI side of things, the process of training the model was difficult as the model was not learning from the training sets properly.

Accomplishments that we're proud of

This project tested our groups ability to think critically and develop a piece of software that has potential. From the web side of things, we were pleased with how our UI came out as it works functionally and doesn't look terrible. On the AI side, inevitably creating a model that learns was a major triumph for our group.

What we learned

There was much to learn from this experience at BrickHack 11. Everyone in our group learned something new about python within this hack. From using data science libraries (i.e. TensorFlow, seaborn, pandas) to backend web frameworks (Django), we all learned something new this weekend about python.

What's next for leafZer0

leafZer0 was prototyped during this hackathon and is of a very primitive form. The next steps for this application are to make it accessible to others by deploying it. But before, we must address some of our infrastructure and data pipeline issues we discovered while hacking. Most notably, the AI can be improved to include sentiment analysis based on local laws and regulations of the dataset, general accuracy, and optimization of the model for speed. From the website, improvements that can be made are a more intuitive UI/UX design, more lenience over data that is passed in, and potentially code migration to a more user friendly front-end framework (i.e. Vue, React).