Project Deep Dives¶
These pages bring project-level details directly into the docs site so students can see concrete commands, outputs, and interpretation patterns without leaving MkDocs.
Included Deep Dives¶
| Deep Dive | Focus | Time | Primary Artifacts |
|---|---|---|---|
| Supervised Learning | classification/regression foundations, imbalance handling, model evaluation | 90-150 min | metrics tables, curves, learning diagnostics |
| Causal Inference | ATE/CATE/tau(x), uplift modeling, targeting policies |
120-180 min | Qini curves, uplift-at-k, policy simulations |
| MLOps Drift | training, drift detection, retrain decisions, local serving | 90-150 min | drift report, policy decision JSON, API outputs |
| Ranking Track | grouped ranking modeling + API productization workflow | 90-150 min | group split manifests, NDCG metrics, ranking API outputs |
| Forecasting Track | time-aware forecasting + observability-ready API workflow | 90-150 min | time split manifests, forecast metrics, demand API metrics |
How To Use These Pages¶
- Pick one deep dive and run the quickstart exactly once.
- Validate artifact generation.
- Use the "How to interpret" checklist to turn outputs into decisions.
- Move to the next deep dive only after you can explain current outputs in plain language.
Source Project Documents¶
- Supervised project source:
projects/sota-supervised-learning-showcase/README.md - Causal project source:
projects/causalml-kaggle-showcase/README.md - MLOps project source:
projects/mlops-drift-production-showcase/README.md - Ranking foundations source:
projects/learning-to-rank-foundations-showcase/README.md - Ranking API source:
projects/ranking-api-productization-showcase/README.md - Forecasting foundations source:
projects/nyc-demand-forecasting-foundations-showcase/README.md - Demand API source:
projects/demand-api-observability-showcase/README.md