Our idea originated from the bushfire crisis in Australia. The number of forest fires in the world is increasing and the destruction it causes is immense. It endangers the lives and livelihoods of local communities.
What it does
The python code uses machine learning to train a model on historic data of input factors such as wind, temperature, drought moisture code etc. that may or may not have resulted in a fire. The code can then predict if there will there be a fire or not based on a new unique combination of these input factors. The web app is used to demo the prediction. The user enters various values and can choose which model he wants a prediction from. We have trained five models which are, Linear Regression, RANSAC, Random Forest, Support Vector Regression and Stochastic Gradient Descent.
How I built it
Using Python, HTML, CSS.
Challenges I ran into
Integrating the web app to the model.
Accomplishments that I'm proud of
Learning how to use HTML and CSS and integrating Machine Learning Models into the web app.
What I learned
Lots of tricks in coding + developed my skills in Python, HTML and CSS.
What's next for Prediction of Forest Fires
As of now the RMSE is relatively high, we believe this is due to over-fitting of data. We aim to sort this out in the next version of the same.