Foundation capstone·5-7 hours

Guided Project: Palmer Penguins

Use all 344 Palmer Penguins observations to complete an auditable pandas and scikit-learn workflow from data contract to model card.

Capstone workspace open from day one

Start with the first project section now. More sections unlock as you complete the preceding labs and studios: 1/5 evidence sections currently open.

Scenario

A field-research teaching team wants a reproducible species classifier as a worked example. You must audit the licensed dataset, preserve an untouched test set, compare against a dummy baseline, cross-validate a leakage-safe pipeline, analyze errors, and state where the model should not be used.

You will demonstrate

Profile a real dataset and define its prediction contract.
Build preprocessing and models inside leakage-safe scikit-learn pipelines.
Compare cross-validation evidence, held-out errors, and limitations in a model card.

Project kit

Palmer Penguins CSVThe complete local 344-observation dataset.Guided project notebookRunnable EDA, pipeline, evaluation, error-analysis, and model-card scaffold.Dataset attributionLicense, citation, and source record.

Project evidence

Show the work, not a checked box.

Each response is stored in this browser as you type. Include metrics, test output, or a decision rationale wherever the deliverable asks for it.

Audit the full dataset

Load the bundled CSV, define one row and the target, report shape, missingness, duplicates, class counts, units, and plausible range checks.

Required evidence: A data-quality table that accounts for all 344 observations and names every excluded row.

Your evidence

0/80 minimum characters

Build a leakage-safe pipeline

Split once with stratification. Fit imputers, encoders, scalers, and the estimator only through a scikit-learn Pipeline and ColumnTransformer.

Required evidence: Pipeline representation, split sizes, and a short explanation of why preprocessing cannot see test rows.

Complete more preceding labs or studios to unlock this section.

Compare baseline and candidate

Report DummyClassifier and candidate cross-validation scores using identical folds, then evaluate the selected workflow once on the untouched test set.

Required evidence: A fold-level metric table with mean, standard deviation, and held-out confusion matrix.

Complete more preceding labs or studios to unlock this section.

Analyze errors and slices

Inspect misclassified rows and performance by island and sex where sample sizes permit. Distinguish data ambiguity from model limitations.

Required evidence: An error table plus two evidence-backed observations and one uncertainty caveat.

Complete more preceding labs or studios to unlock this section.

Write the model card

Document intended use, non-goals, data provenance, evaluation, limitations, ethical considerations, and a next experiment.

Required evidence: A bounded decision memo and complete model-card section.

Complete more preceding labs or studios to unlock this section.

Runnable Python lab

Audit the full Palmer Penguins table

Load the bundled 344-row CSV and implement a compact data-quality summary before modeling.

Dataset: Palmer Penguins, 344 observations

Console

Run the code to inspect its output. Use Check solution only when you want to run the exercise checks.

Code runs only in a time-limited browser worker. It is not sent to the IntuitiveAI server. Do not paste secrets.

Expected result

344 rows, 8 columns, 19 missing cells
Adelie 152, Gentoo 124, Chinstrap 68
ALL TESTS PASSED

Project defense

Which workflow preserves the final test set most effectively?

Rubric self-review

Rate the evidence, not your effort: 0 missing, 1 weak, 2 adequate, 3 strong. All criteria must be reviewed, but a low honest score does not get hidden.

The audit reconciles row counts, missingness, exclusions, target balance, and feature meaning.

All learned preprocessing occurs inside the evaluated pipeline.

Baseline, cross-validation, and final test evidence have clearly separated roles.

Error analysis names concrete rows or slices and respects sample-size uncertainty.

The model card limits use to the evidence and proposes a falsifiable next experiment.

Artifact

Attach the portfolio evidence.

Submit the completed notebook or an exported Markdown/PDF report. The project never executes submitted notebooks.

Validate locallyUpload privately

Anonymous mode stores only metadata and a fingerprint locally. Private upload requires an account and an explicit file selection. Uploaded notebooks are never executed.

Useful references

Palmer PenguinsDataset documentation, variables, provenance, and examples from the project authors.scikit-learn PipelinesOfficial guidance for chaining preprocessing and estimators safely.Model Cards for Model ReportingA reporting framework for intended use, evaluation context, limitations, and ethical considerations.

Project completion gate

Completion is controlled by stored evidence, deterministic tests, a decision defense, rubric review, and the artifact when required.

Evidence pendingCode pendingDefense pendingRubric pendingArtifact pending

Optional cloud portfolio

Submit evidence across devices.

An account is required. Submit only when the local completion gate passes. AI review is advisory and separate from deterministic completion.

After submission, I may choose to send the stored evidence text to the configured AI model for advisory rubric feedback. Artifact bytes are not sent by this review action.

Account settings