Compare spread, not just averages

The question

Compare the centre and spread of delivery durations across depots.

df Spread comparison

Your inputs

`df`

One row is a completed delivery. Durations include unknown values and a genuine long delivery; do not discard an observation merely because it is unusual.

Field meanings and units

delivery_id: Delivery identifier.
depot: Depot responsible for the delivery.
status: Recorded delivery state.
minutes: Elapsed duration in minutes; unknown is not zero.
weight_kg: Weight in kilograms.
date: Calendar date supplied as YYYY-MM-DD text unless the task states it is invalid.

`df` · first 5 of 24 rows
`delivery_id`	`depot`	`status`	`minutes`	`weight_kg`	`date`
D001	East	Delivered	32	1.2	2026-04-01
D002	West	Delivered	48	1.9	2026-04-01
D003	North	Delivered	25	2.6	2026-04-02
D004	East	Delivered	None	3.3	2026-04-02
D005	West	Delivered	45	4	2026-04-03

View all 24 rows

`df` · complete input
`delivery_id`	`depot`	`status`	`minutes`	`weight_kg`	`date`
D001	East	Delivered	32	1.2	2026-04-01
D002	West	Delivered	48	1.9	2026-04-01
D003	North	Delivered	25	2.6	2026-04-02
D004	East	Delivered	None	3.3	2026-04-02
D005	West	Delivered	45	4	2026-04-03
D006	North	Delivered	37	1.2	2026-04-03
D007	East	Delivered	72	1.9	2026-04-04
D008	West	Delivered	29	2.6	2026-04-04
D009	North	Delivered	bad	3.3	2026-04-05
D010	East	Delivered	49	4	2026-04-05
D011	West	Delivered	26	1.2	2026-04-06
D012	North	Delivered	62	1.9	2026-04-06
D013	East	Delivered	46	2.6	2026-04-07
D014	West	Delivered	38	3.3	2026-04-07
D015	North	Delivered	73	4	2026-04-08
D016	East	Delivered	30	1.2	2026-04-08
D017	West	Delivered	34	1.9	2026-04-09
D018	North	Delivered	50	2.6	2026-04-09
D019	East	Delivered	27	3.3	2026-04-10
D020	West	Delivered	63	4	2026-04-10
D021	North	Delivered	47	1.2	2026-04-11
D022	East	Delivered	180	1.9	2026-04-11
D023	West	Delivered	74	2.6	2026-04-12
D024	North	Delivered	31	3.3	2026-04-12

The editable setup in Your Python creates these inputs when you run it.

Your task · Compare spread, not just averages

Compare the centre and spread of delivery durations across depots.

observationsDataFramefigFigure

Deliverables

Make the outputs listed below. Use each name exactly as shown. Then choose Check answer.

Python name key

Teal outlines name table columns. Violet left rules name Python objects. Dotted rules mark syntax, functions and parameters. Use each name exactly as shown.

Valid delivery measurementsOutputobservationsDataFrame
1. Create observations with depot and parsed minutes for known durations, keeping source order.
FigureOutputfigFigure
1. Create fig comparing depot medians, middle 50%, and 1.5-IQR whiskers, including outliers.
2. Put depots on the horizontal axis and minutes on the vertical axis.
3. Add a meaningful title and labels
4. Display the figure.
Identify Depot and Minutes on the appropriate axes; extra wording is welcome.

Data policy Keep all supplied source tables unchanged. Work on separate outputs.

Help Hints & explained solution

Choose the amount of help you need. Opening one hint does not reveal the others.

Hint 1 — Think

What could two groups with similar means still differ in?

Hint 2 — Tools

to_numeric(), dropna(), sns.boxplot() or Axes.boxplot()

Revisit: Convert numeric text · Drop missing rows · Box plot · Finish a chart properly

Hint 3 — Approach

Retain known numeric durations and compare each depot’s median, middle half and whiskers using the standard 1.5-IQR rule.

Explained solution

Add these lines below # Your work in the editor. The supplied input setup is already there.

work = df.copy()
work['minutes'] = pd.to_numeric(work.minutes, errors='coerce')
observations = work[['depot', 'minutes']].dropna()
fig, ax = plt.subplots()
sns.boxplot(
    data=observations,
    x='depot',
    y='minutes',
    ax=ax,
)
ax.set(title='Delivery duration spread', xlabel='Depot', ylabel='Minutes')
fig.tight_layout()
plt.show()

View supplied setup

This code is already at the top of the editor.

import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns

# Supplied input: df (24 rows)
df = pd.DataFrame(
    [
        ('D001', 'East', 'Delivered', 32, 1.2, '2026-04-01'),
        ('D002', 'West', 'Delivered', 48, 1.9, '2026-04-01'),
        ('D003', 'North', 'Delivered', 25, 2.6, '2026-04-02'),
        ('D004', 'East', 'Delivered', None, 3.3, '2026-04-02'),
        ('D005', 'West', 'Delivered', 45, 4, '2026-04-03'),
        ('D006', 'North', 'Delivered', 37, 1.2, '2026-04-03'),
        ('D007', 'East', 'Delivered', 72, 1.9, '2026-04-04'),
        ('D008', 'West', 'Delivered', 29, 2.6, '2026-04-04'),
        ('D009', 'North', 'Delivered', 'bad', 3.3, '2026-04-05'),
        ('D010', 'East', 'Delivered', 49, 4, '2026-04-05'),
        ('D011', 'West', 'Delivered', 26, 1.2, '2026-04-06'),
        ('D012', 'North', 'Delivered', 62, 1.9, '2026-04-06'),
        ('D013', 'East', 'Delivered', 46, 2.6, '2026-04-07'),
        ('D014', 'West', 'Delivered', 38, 3.3, '2026-04-07'),
        ('D015', 'North', 'Delivered', 73, 4, '2026-04-08'),
        ('D016', 'East', 'Delivered', 30, 1.2, '2026-04-08'),
        ('D017', 'West', 'Delivered', 34, 1.9, '2026-04-09'),
        ('D018', 'North', 'Delivered', 50, 2.6, '2026-04-09'),
        ('D019', 'East', 'Delivered', 27, 3.3, '2026-04-10'),
        ('D020', 'West', 'Delivered', 63, 4, '2026-04-10'),
        ('D021', 'North', 'Delivered', 47, 1.2, '2026-04-11'),
        ('D022', 'East', 'Delivered', 180, 1.9, '2026-04-11'),
        ('D023', 'West', 'Delivered', 74, 2.6, '2026-04-12'),
        ('D024', 'North', 'Delivered', 31, 3.3, '2026-04-12'),
    ],
    columns=['delivery_id', 'depot', 'status', 'minutes', 'weight_kg', 'date'],
)

Why these steps exist

The middle half and whiskers describe spread that a single average hides.
The long delivery remains visible as an outlier under the 1.5-IQR rule; unusual is not the same as invalid.

Another valid approach

Matplotlib or Seaborn box plots are accepted with the same group quartiles, 1.5-IQR whiskers and visible outliers.