0
$\begingroup$

I have got a binary classification problem with large dataset

Also dataset is fairly balanced of 67% Class 0 , 33% Class 1.

My accuracy score is very less in test dataset as compared to train dataset which is clearly a case of Overfitting

enter image description here

Also my classification report is

enter image description here

How to improve my model? I have tried undersampling , cross validation , lasso feature selection , etc.

$\endgroup$
2
  • 1
    $\begingroup$ The results are already much better than last time, congrats! We would need more detail to help you: what kind of task are you doing? how many instances? how many features? which learning algorithm? $\endgroup$
    – Erwan
    Mar 20, 2022 at 15:58
  • $\begingroup$ I have used random forest classifiers , extra tree classifiers for prediction. There are 55 features in my dataset with 1 mill records. $\endgroup$
    – Shubh
    Mar 21, 2022 at 2:00

1 Answer 1

2
$\begingroup$

I would empasize on more feature engineering like create more derived transformed features (log,interaction features (additive,division) etc.), trying out different bins, one-hot encoding, play with features based on business understanding etc.

$\endgroup$
1
  • $\begingroup$ Upvoted, very helpful keep up the good work... $\endgroup$ Mar 30, 2022 at 5:18

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.