Tuning parameters for LGBM models - KOBE

[mfzeidan]

Kobe

below is a workbook I've put together from kaggle. There is a section in this code (second box) that has a list of parameters. Messing with these numbers (changing them however you'd like) will help improve or worsen the model. We'd like you to spend some time going through this page and tuning the parameters and noting which helped improve the mode.

http://192.241.145.188:8888/notebooks/Marks%20stuff/lgbm_param_tuning.ipynb

let me know if you have any questions, we're glad to help :+1:

note the comment at the bottom of the last box!!! -

• after you run this box (shift + enter) you'll see this character '' next to the In[] box at the front.
• once this changes to a number, the code is complete. do not run the code below until the * character is gone