ML Process

1. Acquire labelled or annotated data
   • Gold Standard Data
2. Consider Ethical Considerations in ML
3. Perform Data Cleaning
4. Validation in small datasets
   • Cross-validation
5. Analyze Data and features using visualizations, dimension reduction
6. Perform Feature Engineering
7. Shortlist models
   • pick complexity of the model for the perfect bias-variance tradeoff
8. Fine tune the model
   • set Hyperparameters
     • data transformations & feature engineering are hyperparameters too
     • perform random search, grid search
     • Automatic hyper parameterization using Auto-ML
9. Fit on training + validation data, test on test data
10. Pick best performing model or aggregate models via Ensemble methods
11. Evaluation of ML systems