R19 · Week 19 Recall & Drill

Weekly revision · Week 19 · Covers 5 sessions from Mon–Fri.

Sessions covered

• S091 — Gradient Boosting — XGBoost, LightGBM
• S092 — Evaluation Metrics — P/R/F1/ROC/PR/AUC
• S093 — Feature Engineering — Encoding, Scaling, Missing
• S094 — Imbalanced Data — SMOTE, Class Weights, Thresholds
• S095 — Model Selection — CV, Hyperparameter Tuning, Optuna

---

90-min structure

1. Warm-up recall · 5 min

Close all tabs. Write down (on paper or a text file) everything you remember about each of the 5 sessions above. No peeking.

2. Blank-page reconstruction · 30 min

For each of the 5 sessions:

• Draw the key diagram / architecture from memory
• State the one-sentence intuition
• List 2 gotchas or "production realities"

Compare with the original session. Note the gaps.

3. Hands-on drill · 30 min

Pick one session's hands-on code from this week. Re-implement it from scratch without looking at the original. Then diff and understand where you went wrong.

4. Anki cards + spaced repetition · 15 min

Create 3–5 flashcards per session (~15–25 cards for the week). Format:

• Q: "What is X?" → A: one sentence.
• Q: "When would you use X?" → A: one concrete example.

5. Gap analysis + next-week preview · 10 min

• What did you get wrong in step 2?
• What was hardest in step 3?
• Skim the titles of next week's 5 sessions — set expectations.

---

Bonus challenge

If this week's sessions clustered around one theme, try to write a 200-word blog post explaining "this week I learned…" — teach it back publicly.

---

Part of the 6-month evergreen learning plan.