crossword puzzle

2. Differences between predicted and actual outputs in regression analysis.
3. A clustering metric showing how well a point fits within its cluster.
6. A value set before training that governs model behavior and must be tuned for optimal performance.
13. Technique used repeatedly across the syllabus for model validation and robustness.
14. The phenomenon where model performance declines as real-world data distribution changes.
16. Process of adjusting hyperparameters to improve model performance.
17. Measure of similarity between a time series and its lagged version.
18. A crucial PA consideration regarding privacy, fairness, and responsible model use.
19. An operator in evolutionary algorithms introducing random variation.
20. A mathematical or ML structure trained to learn patterns.
22. In genetic algorithms, a measure of solution quality.
24. A general term for a model’s output or evaluation measure (classification or regression).
26. A regression error metric penalizing larger deviations more strongly.
28. A resampling method to prevent overfitting while estimating a model’s generalization ability.

1. An unsupervised task grouping data based on similarity.
3. You only unlock this answer once the grid around it is complete.
4. Indicates whether predictions systematically overestimate or underestimate true values.
5. A time series model combining autoregression, differencing, and moving average.
7. A dataset issue where one class significantly outnumbers others, harming model performance.
8. A regression metric measuring the average magnitude of errors without direction.
9. The process of adjusting parameters or operations to minimize loss or maximize performance.
10. A widely used classification metric plotting sensitivity vs. fall-out.
11. A forecasting metric expressing prediction error as a percentage.
12. A statistical method of sampling with replacement for robust estimation.
15. Neural network architecture used for sequential and time-dependent data.
21. A gated RNN variant that handles long-term dependencies in forecasting.
23. Combines multiple models to improve predictive performance.
25. A predictive modeling technique for estimating numerical outcomes.
27. The process of assessing model performance using metrics such as RMSE, AUC, etc.