FireWatch

Inspiration

Our team's inspiration for the project came from us growing up in California surrounded by wildfires. Given our interest in machine learning and app development, it seemed natural to integrate the two in a meaningful way to help prevent wildfires.

What it does

Our app allows Californians to easily access information on fire risks and ways to be fire-safe and prevent wildfires. This app delivers information to the public in an effort to educate the public and prevent their state from continuing to burn.

How we built it

The app UI was built with Figma while the machine learning model we designed was implemented primarily in GNU Octave (essentially open-source MATLAB).

Challenges we ran into

The primary challenges we ran into were time and knowledge constraints. Given that this was our first hackathon and we had a lack of people, we were at a disadvantage when it came to our ideas being deployed into our final product. Our brainstorming session and project outline details using cloud servers, stochastic gradient descent, APIs, and more in order to accomplish our goal, but a lack of data and knowledge with cloud-based computing meant we couldn't accomplish some of our objectives. Nevertheless, with the constraints given, we managed to create a functioning model that somewhat accurately predicts the fire risk based on temperature, humidity, and other weather factors.

Accomplishments that we're proud of

We are proud of having fully created a project outline that could eventually become an incredible initiative given more development in the future. In addition, our deployment of the app interface and machine learning model meant that we had accomplished the two main objectives we set for ourselves. Huge pats on the back for ourselves on that.

What we learned

The biggest thing we learned was what we haven't learned: our knowledge gaps really shone through when we tried putting some of our ideas into practice, and as such we are going to develop our skills in the future such that we won't face the same problems. Additionally, we learned a lot about development in a more professional and fast-paced environment, one that more realistically resembles a work environment as opposed to the more casual teams we are used to.

What's next for FireWatch

FireWatch started off with an incredible project outline that was somewhat overambitious given the time and resource constraints we dealt with. That being said, we would love to continue development on the project and accomplish some of the ideas we had in mind such as (but not limited to): finding better datasets for more feature-rich and accurate results; implementing a cloud-based storage and computing system to relocate our datasets and model calculations off of local machines; implementing APIs, stochastic gradient descent, and a time-based weighting of data for more up-to-date and accurate predictions.

Built With

• css
• figma
• octave