Predicting Wildfire with AI

Using weather API data online and wildfire reports from NASA, we have designed an AI that allow us to predict the probability of wildfire when given the data like wind, humidity, precipitation, etc.

Comment 2

Inspiration

To reduce the impact of natural disasters across the globe using AI technologies and data collected by NASA/Vision crossing APIs. Specifically, our goals have been to minimise the looses and causalities resulted from wildfires by detecting the risk better and allowing for earlier intervention.

What it does

This project uses the data of temperature, humidity, wind speed, and other data to predict the probability of wildfire. Not only does it calculate for the likelihood of a fire at a region, but it also predicts when it would occur under certain conditions. These calculations allow for earlier notifications and interventions, allowing us to mitigate the damage wildfires bring

How we built it

We find the data from the website and put the data set into Google Collab to make the prediction.

Challenges we ran into

The longer the time period of the data set we study, the higher the accuracy is. Therefore, we chose the data from 2000 to 2024, which is a huge sample, this is difficult for us to find a suitable and credible data set.

Accomplishments that we're proud of

As an inexperienced team, this project gave us a lot of sense of achievement. Although we lack expertise in fields such as data analytics, artificial intelligence, etc., we were able to overcome many challenges and make significant progress, which gives us the confidence to face other problems in the future.

What we learned

As first time participants in Hackathons, we have learned the importance of problem solving, time management, and collaboration. On this journey, it has been meaningful and fulfilling to expand our comfort zones, learn new things, and meet with other promising engineers in the field.

What's next for Predicting Wildfire with AI

We need to further train the program to make the prediction more accurate. Although so far we already have a simple trial, the nature is complex, which needs to involve more advanced machine learning algorithms.

Built With

Share this project:

Updates

posted an update —

This code uses the pandas library to load a weather dataset and print some initial data like column names and sample rows. It integrates with Google's Generative AI to analyze the dataset by sending small summaries of the data to the AI model. Several analyses are performed, like summarizing the dataset, identifying common weather conditions, and calculating the correlation between temperature and humidity. To avoid overwhelming the AI, large datasets are truncated, and exceptions are handled to ensure smooth execution if something goes wrong. Finally, results for each analysis are printed.

Log in or sign up for Devpost to join the conversation.