ta-training

Interview Exercise: 5-Minute Code Review

This exercise simulates what TAs will do with students in the Data Science Clinic:
review real code, identify high-value improvements, and give constructive feedback.

---

Task

• Skim the script bad_script.py.
• Propose 3–5 concrete suggestions to improve correctness, clarity, or maintainability.
• For each suggestion, briefly explain what to change and why.
• Do not run the code or rewrite it in full — this is a review exercise.
• Assume Python 3.10+ and pandas are available.

Time limit: 5 minutes.

---

What We’re Looking For

High-value points (examples below — you don’t need all of them, just 3–5 solid picks):

• Add a __main__ guard so imports don’t execute I/O.
• Replace print with logging at appropriate levels.
• Avoid mutable default arguments (e.g., groups=[]).
• Add type hints and docstrings for functions.
• Separate computation from I/O (functions return values, wrappers handle file writes).
• Don’t use broad except Exception; catch specific errors.
• Avoid inplace=True and chained assignment pitfalls.
• Validate required columns explicitly.
• Pass file paths and parameters as arguments instead of globals.
• Use Pathlib for file paths.
• Write JSON with DataFrame.to_json instead of json.dumps.

---

Rubric (for graders)

• 1 point each for a correct, high-value observation with rationale.
• Max: 5 points.
• Extra credit: clear explanation of why the change matters.

---

Files

• bad_script.py — intentionally flawed version.
• reference_solution.py — revised version illustrating best practices.