Here are the notebooks and datasets that the Undergraduate Artificial Intelligence Society at the University of Alberta used to validate the submissions for the competition, UAIS September 2023 Classifying Alberta Wildfires on Kaggle. Due to a shift in the metrics used to evaluate the submissions on Kaggle the final leaderboard is not representative of the true scores for each team. As a result, we decided to check each submission against the updated final_solution.csv using a weighted f1-score to evaluate the results. Here is the updated and final leaderboard:

Congrats to the winners:

1. Yong W. Lee
2. Alexander Bradley
3. aman7999

Thanks to everyone who participated!

If you want to see the process that went into finalizing the submissions view the 2023_September_Datathon_Submissions notebook.

A problem that we ran into, was in the process of creating the solution file that was on Kaggle some of the size classes got scrambled. So the solution file that is on Kaggle is not the one that we used to validate the results. Therefore we had to create a true solutions file by merging the test.csv and the original wildfires dataset. This process is shown in the submissions notebook above. If you want to learn more about the process that went into making the wildfire dataset from the dataset from the Alberta Government, look here.

If you are wondering which submission file we used for each team, we used the submission that was flagged by the team for the final score as evidenced by the flag in the photo below:

If you have any questions feel free to reach out to any of the executives on Discord or send us an email at [email protected]. Thanks again to everyone who participated, we wish you an awesome rest of the semester!