404 not found -IM38

Fire Classification
Density of Wildfires in California
Fire Classes

Inspiration

We wanted our project to be grounded in a real world problem. Having seen the increased frequency in which wildfires appear in the news, and the impact they have both environmentally and economically, we thought this would make for an interesting project!

What it does

Our project uses machine learning to classify areas that are at risk of wildfires, and also predicting how large these wildfires could grow. We also performed exploratory data analysis and made visualisations to better convey our results

How we built it

We built our solution using the IBM-Z platform. All of our code is written in python or MySQL. Various machine learning and imaging libraries were used.

Challenges we ran into

One major challenge we ran into was our linux server shutting down for an hour or so after lunch, which forcced us to code locally whilst simultaneously figuring out a way to get it back online. Another was building our own comprehensive dataset from multiple smaller datasets from various sources.

Accomplishments that we're proud of

-Building our own comprehensive dataset -Solving the linux server crashing problem -Increasing a models performance by 13% by downsampling a dataset and using dropout