Alberta Wildfire Prediction

What it does

We examine Alberta's wildfires from 2006-2021 under a data science perspective. We provide useful insights on wildfire trends and an XGBoost prediction model.

How we built it

We began by preprocessing the dataset extensively; this included normalizing values, imputing data, and feature engineering. We then visualized the data to uncover patterns and gain intuition about the problem. After, we developed an XGBoost model to predict wildfire size classes with ~90+% accuracy. Finally, we summarized our findings and methodologies in a paper.

Challenges we ran into

We ran into many challenges along the way, from preprocessing missing values and removing outliers, identifying important features, and tuning the model, among many others!

Accomplishments that we're proud of

This was the first data science experience for the entire team; we learned Pandas and Matplot, various machine learning models like Random Forests and LightGBM, to best practices like hyperparmeter-tuning and encoding strategies!