BitWithBite
AI & Machine Learning Worksheet
🤖 Building a Full Pipeline
SCORE___ / 20
NameClassDate
After this worksheet you can
ColumnTransformer for mixed dataCombine numeric + categoricalFit once, predict anywhereGrid-search a pipeline📚 Quick Recap
You've seen estimators, transformers, and basic pipelines. Now you'll build one that handles a realistic dataset — mixed numeric and categorical columns, missing values, and a final model — all wrapped in a single object you could deploy tomorrow.
🧠 Section A · Concept Check ● BEGINNER 4 × 1 = 4
1Different preprocessing per column type uses:
2A pipeline guarantees test data gets:
3Pipeline step params are tuned as:
4pipe.predict runs preprocessing:
🧮 Section B · Problem Solving ● INTERMEDIATE 2 + 3×3 = 11
5Transformer routes columns to the right prep.
6Numeric columns get a scaler; categorical get an .
7Sketch a ColumnTransformer for [age, salary] + [city].
8Why grid-search the whole pipeline, not just the model?
9What breaks if you scale before splitting?
🚀 Section C · Challenge ● CHALLENGE 5
10Name the two big wins of pipelines.
💭 Reflection — the most useful thing I learned:
A ___/4 B ___/11 C ___/5 Total ___/20
Teacher's Signature
Parent's Signature
✂ answer key — fold or cut before handing out
1-A 2-A 3-A 4-A | 5 Column 6 encoder 7 = ColumnTransformer([('num', StandardScaler(), ['age','salary']), ('cat', OneHotEncoder(), ['city'])]) 8 = CV folds re-fit preprocessing per split - no leakage 9 = Test stats leak into training - overly optimistic scores | 10 = No leakage + reproducible single object to deploy