Inspiration

  1. Unpredictability of wildfire: many current news sources such as ca.fire.gov and latimes.com currently show recent news about wildfires, which often comes too late for California residents to evacuate. Thus, we need a system that can predict wildfire movements, impact and spreading speed
  2. Lack of knowledge of wildfire severity: current wildfire news sources do not provide clear information about the severity of the fire, so users often do not know whether they need to evacuate or not.

What it does

Our application uses EazyML's model built on training data found on Kaggle's California wildfire dataset ranging from 2013-2019 in order to predict the likelihood and severity of wildfires in California. It also shows the timeline of wildfire severity across California, and the greatest causes for wildfire severity: acres burned, longitude, latitude, and crews involved.

How we built it

Frontend: Bootstrap for HTML, JS, CSS Backend: Flask, EazyML, Folium, NewsAPI Dashboard: From Creative Tim's Argon Dashboard Flask

Challenges we ran into

EazyML's API calls are different from the normal sklearn or Keras, but after reading the clear documentation, the process was clear.

What we learned

We learned how to pair EazyML with Flask. It was also a good collaborative coding experience, which taught us about the importance of prioritizing tasks, managing time, and modularizing jobs.

What's next for Extinguish

We plan on licensing our application to government officials, which will allow us to provide the application free of cost to all California residents. We also plan to make a mobile app to enable easy access.

Built With

Share this project:

Updates