FireCheck

With FireCheck, you will know if you are in danger from any fires up to 15 days in advance by checking the probability of fires before they begin- allowing you to pack necessities and get to safety!

Comment

Inspiration

Inspired and saddened by the devastating recent wildfires around Los Angeles, we realize it is important to be able to take the necessary precautions to stay safe from any wildfires in order to prevent the loss of lives and damage to property. There may not be easy access of information for people to be aware of nearby wildfires.

What it does

This website allows users to select their location of interest and check the real-time probability of wildfires happening. There is a chatbot function for users to ask questions regarding wildfires, and get tips about how to stay safe.

How we built it

  1. We used weather data from the California Irrigation Management Information System, which provided us with hourly information on California weather. We used data from six different counties and combined with a dataset of the last 100 wildfires in California. We performed a logistic regression model in Python to determine the weight of different weather parameters. By inputting each of the current weather parameters into the model, we can obtain the probability of wildfires in real time.
  2. For the Gemini Chatbot, we used the API visual crossing and utilized Streamlit to visualize data. We used Flask, JSON, and Python to run the Google generative AI.
  3. We produced the front-end webpage using HTML, CSS, and JavaScript.

Challenges we ran into

  1. It was difficult to find accurate weather data. There were very few large enough datasets and most datasets available have a limited range of data, leading to biased information as there were missing parameters.
  2. There were little datasets on past fires available but we were able to find one from a government website. We combined this dataset with the weather data to determine the weights of each parameter that influences the probability of a wildfire.
  3. We encountered challenges when creating the chatbot, including technical errors while integrating the chatbot into the website with the HTML file.
  4. The equation used to calculate the probability was initially inaccurate. We had to undergo multiple modifications to increase the accuracy of the equation by incorporating more parameters.

Accomplishments that we're proud of

We were able to integrate the chatbot to the website using Streamlit. We also designed a webpage that has a dropdown box for users to select their location of interest easily to check the wildfire probability there. We were able to calculate the probability and display each parameter on the webpage as well.

What we learned

Through AthenaHacks 2025, we have learned how to utilize some tools and libraries we did not have previous experience with, including Streamlit, Gemini, and Figma, as well as employing the logistical regression model to perform calculations. In addition, we have understood the importance of collaborative teamwork and communication between each team member while working on this project!

What's next for FireCheck

Expand the fire predictions to more areas outside of California - ideally including coverage across North America, since many parts of Canada are also prone to frequent wildfires! Also, we can continue incorporating more parameters and use larger datasets to increase the accuracy of our predictions.

Attempted Prizes

Best Overall Hack, Athena’s Favorite Hack, Best Sustainability Hack, [MLH] Best Use of Streamlit, [MLH] Best Use of Gemini API, [MLH] Best Use of AI powered by Reach Capital

Team Members

Discord Usernames: shriya.r, tiffanyk0, purpleyams, suminwu

Built With

Share this project:

Updates