Skip to content

Notebook for the Fire fighting using data on Zindi. Ranked number 5 on the public leaderboard and 8 on the private leaderboard. https://zindi.africa/hackathons/cmu-africa-fighting-fire-with-data

Notifications You must be signed in to change notification settings

boladjivinny/fire-prediction

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

15 Commits
 
 
 
 
 
 
 
 

Repository files navigation

CMU-Africa: Fighting Fire with Data

Competition held on Oct. 31, 2020

Brief summary

The task consisted in predicting the percentage of area burnt for over specific periods of month using the dataset provided. To solve the regression problem, I essentially did some feature engineering, combining some features together and applying functions to others to draw more insights (exp, log, x^2, etc.). By proceeding like that, I was able to get a RMSE of 0.23 on the public leaderboard with the RidgeCV algorithm.

A series of boosting algorithms have then been tried and then stacked with the linear regression model to achieve the final position on the leaderboard.

A few more feature engineering would certainly have yielded positive results along with hyperparameter tuning that I could not finish on time due to time constraints.

About

Notebook for the Fire fighting using data on Zindi. Ranked number 5 on the public leaderboard and 8 on the private leaderboard. https://zindi.africa/hackathons/cmu-africa-fighting-fire-with-data

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published