Skip to content

Lab 03 — Jupyter in Depth

Session: Week 2 · Session 3 · 90 min CLO: CLO-1 Difficulty: Beginner

Learning objectives

By the end of this lab you can:

  1. Use markdown cells with headings, lists, tables, and code formatting.
  2. Use magic commands (%timeit, %matplotlib inline, %%time).
  3. Explain what a kernel is and restart/clear outputs safely.
  4. Keep a notebook clean: run top-to-bottom, no hidden state.

Problem statement

A messy notebook is a liability: a cell that "worked earlier" fails after a restart because it depended on hidden state. You will build a clean, restart-safe notebook that analyzes a small dataset, and you will prove its cleanliness by restarting the kernel and running all cells in order.

Dataset requirements

Seaborn's built-in tips dataset (244 rows). Load it with sns.load_dataset("tips") — no download needed.

Step-by-step tasks

  1. Title + intro cell: a markdown cell with a # heading, a one-line description, and your name/date.
  2. Imports cell: put ALL imports in the first code cell.
  3. Data load cell: load tips and print shape and head().
  4. Analysis cell: compute the mean tip by day using groupby, show it as a markdown table in the NEXT (markdown) cell.
  5. Speed cell: time a cell that computes df["tip"].mean() with %timeit — write down the number in markdown.
  6. Prove restart-safety: use the Jupyter menu Kernel → Restart & Run All. Every cell must run top-to-bottom with no errors.
  7. Trap check: write a code cell that uses a variable secret which is NOT defined in any earlier cell. Restart & Run All — the cell must fail. Fix the notebook by defining secret properly, then re-run cleanly.
  8. Hygiene pass: clear all outputs (Edit → Clear All Outputs), then Restart & Run All one final time.

Starter code

# Cell 1 — imports (everything in one place)
import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
%matplotlib inline

# Cell 2 — load data
tips = sns.load_dataset("tips")
print("shape:", tips.shape)
tips.head()

# Cell 3 — analysis
mean_tip_by_day = tips.groupby("day")["tip"].mean().round(2)
mean_tip_by_day

# Cell 4 — timing
%timeit tips["tip"].mean()

# Cell 5 — the trap (intentionally broken at first)
print(secret)

Expected output

  • Cell 2: shape: (244, 7) plus the first 5 rows.
  • Cell 3: a Series with mean tips per day (Fri 2.73, Sat 2.99, Sun 3.26, Thur 2.77 approximately).
  • Cell 4: a %timeit result like 12.1 µs ± ... per loop.
  • Cell 5: NameError: name 'secret' is not defined until you fix it; after the fix it prints whatever value you assigned.
  • Final run: zero errors from top to bottom.

Questions

  1. What is a kernel, and what does "restart" actually reset?
  2. Why does the order of cells matter even though Jupyter lets you run them in any order?
  3. What does %timeit do differently from %%time?
  4. Your notebook runs fine, but a classmate's copy errors with NameError. What is the most likely cause?
  5. Why should imports live in the first cell?

Challenge task

Add a markdown "report" cell at the end that summarizes the tips dataset in exactly three sentences: one about size, one about the day with the highest mean tip, and one about the range of total_bill. Format the three sentences as a bulleted list, and bold the day name. Keep it restart-safe.

Marking rubric

Criteria Max Notes
Title/intro markdown cell 2 heading, description, name/date
Imports in one first cell 2 verified in final run
Data load + shape output 2 (244, 7)
Groupby analysis + markdown table 3 numbers match
%timeit result recorded 2 value in markdown
Trap cell fixed properly 3 defined before use, not deleted
Final Restart & Run All clean 3 no errors
Challenge: 3-sentence report 3 bullets, bolded day, restart-safe
Total 20