Thanks to everyone who made this competition possible, Hong Kong University for providing the data, Zindi for hosting it, and to all the competitors for making it more challenging for each one of us.
Don't forget to Leave a⭐️ to my repo, if you find it useful😊
Thanks @Babawale_Sodiq - I note a few things. You counted number of unique geologies, you used gradient multiplier to address unbalanced data (most others I saw used SMOTE or thresholds) and you had a few nice features! You used e.g. ranges and ratios to measure dispersion.
I am really intrigued with what made you able to do so well in this comp and really appreciate that you shared the whole solution! Thanks.
Thank you for sharing your solution. Really great approach. Just a question what feature selection technique did you use to select the features used in modelling?
Congratulations on your first place!
Thanks very much for sharing your solution, it is much appreciated!!