Here is the link to the git repo https://github.com/thusodangersimon/sanral_hack
Thanks for sharing!
I notice you're looking at accidents specifically (based on isAccident) in the EDA. I believe the challenge is to predict 'road incidents' (all incidents in the training file would count) and not just accidents. Makes it slightly less imbalanced.
Thanks for sharing
Oh thanks for letting me know!