Improving ROC_AUC_SCORE of my ML model

#1

The above is the link for the github link for the issue. Currently I am getting 0.85 score on a contest. I want to improve this score. Pls guide me what all changes i can do to achieve the same

Thank you

#2

Good Morning Sir. This is the link for the notebook

#3
#4

hey @samalprathamesh123 ,
there are some changes need to be made .

  1. use the same label encoder for testing. Do not fit it again and again.
  2. For imbalanced data, oversampling can be good, but not always. Use different models to tackle it.
    Like you can try XGBOOST , specifically made to handle unbalanced classes.
  3. Perform model parameter tuning.
  4. Feature engineering is the most important. just checking feature importances is not enough.
  5. use cross validation to evaluate the model working .
  6. Try different tree models, decision tree and random forest doesn’t work good always.

I hope these can help you improve your score.

#5

Got it. Thank you for the tips. I will keep this in mind.

#6

I hope I’ve cleared your doubt. I ask you to please rate your experience here
Your feedback is very important. It helps us improve our platform and hence provide you
the learning experience you deserve.

On the off chance, you still have some questions or not find the answers satisfactory, you may reopen
the doubt.

#7

definitely, i would be happy to do that. Actually I am unable to find a link/button from where I can my amazing experience. I request you to kindly guide me regd. the same

#8

Edit: I am unable to find the rate/feedback option on this portal*

#9

no need to worry then.
just leave it for now.

#10

I got the link and have rated. Thank you so much.

#11

ohhkay.
Thank You so much for that.