My Participation in ML challenge. 1st participation and it was very good learning!
Learnings:
- Use of RFE for feature engineering.
- Trying out different models simultaneously.
- Finally using Bayesian Optimization.
- Feature engineering is the crucial part because hyperparameter tuning can be taken care by Bayesian Optimization.
- SMOTE and oversampling techniques seldom work for unbalanced datasets. Hit rate of <0.5%.