Choosing the Chart

P14.chart-craft.02 · Audience: guest, it-ml, language-pro · Prerequisites: Visual Encodings — What the Eye Ranks Well

Visual Encodings — What the Eye Ranks Well Scales, Axes & Legends

LLM grading (this page):Real gradingOff — no LLM calls

Most chart advice is a catalogue: here is a bar chart, here is a line chart, here is a pie — pick the one that looks right. That is backwards. A chart is not a thing you pick; it is the answer to a question, and if you name the question first the chart is nearly decided for you. "Which region sold most?" wants a comparison. "Is revenue rising?" wants a trend. "Are salaries clustered or spread?" wants a distribution. Each of those questions has a form that serves it well and several that serve it badly — and sometimes the honest answer is that no chart beats a plain table. This module gives you the mapping from task to form, the five task families every business chart falls into, and the underrated case where a table wins. Then a ladder puts it in your hands: given a task and a dataset, choose and defend the form.

🗺️ So far you have built: the perception ranking — position and length are read almost exactly, area and colour only loosely (Visual Encodings) — so you already know which channels a chart form spends. This module uses that to choose forms: every good form puts the reader's key comparison on a high channel.

In this module you will: how to choose a chart form from the task it must serve, not from a gallery. You will meet the five task families — comparison, trend, distribution, part-of-whole, relationship — and the form each one wants; see on four concrete numbers how the same data answers four different questions with four different charts; and learn the test for when a plain table beats every chart. The ladder then has you map tasks to forms, catch a mismatched chart, and justify a table.

Step 1 / 5 — Name the task before you name the chart

Before reaching for any form, finish this sentence: "the reader should be able to see, at a glance, that ...". The end of that sentence is the task, and the task is what the chart exists to serve. A chart that shows the data but not the task is a decoration; a chart chosen for the task is a tool. The mistake the gallery encourages is to start from the shape — "let's do a pie" — and only then ask what it shows. Start from the question and the shape follows.

ⓘ Concept: A chart is the answer to one question — decide the question first

'the reader should see at a glance that ___' \to the blank is the task \to the task picks the form (and the channel the ranking says to use)

Why it matters — Almost every bad chart is a chart built without a stated task: it faithfully renders the numbers but leaves the reader to guess what they were meant to notice. Naming the task first does two things at once. It tells you which comparison must be easy — and from the perception ranking you already know that comparison belongs on position or length. And it tells you which comparisons you are willing to make hard, because a single chart cannot make every comparison easy at once. The discipline is one chart, one question; if you have two questions, you very likely need two charts, or a table.

✅ Key takeaways

A chart is the answer to a question, not a shape you pick from a gallery. Finish 'the reader should see at a glance that ___' first — the blank is the task, and the task decides the form.
Almost every analytics chart serves one of five task families: comparison (bar/dot), trend (line), distribution (histogram/box), part-of-whole (stacked bar; pie only for 2–3 slices), relationship (scatter).
The default forms are defaults because each spends a high perception channel on its task — length for comparison, slope for trend, position×position for relationship. Knowing why lets you judge when to depart from the default.
The same sixteen numbers answer four questions with four different charts; no single form answers all well. The data did not change — the task did, and the task chose the form.
The most underused chart is a table: it wins when the reader needs exact values, when there are only a few numbers, when rows carry mixed units, or when the task is lookup rather than pattern. Match the form to the task, and sometimes the best form is no chart.

Next: P14.chart-craft.03 — Scales, Axes & Legends

Ask the mentor about this module

Ask a question about this content. The mentor explains and grounds its answer in what you are studying; asking is recorded as a learning signal, not a grade.

Ctrl/Cmd + Enter to send

🎓 Practice ladder

4 graded rungs · ~35 min

Now choose forms yourself. Each rung is a three-panel workspace: instructions on the left, a code editor in the middle, output and test results on the right. Run checks the visible tests; Submit grades against hidden cases. The explorer rung maps tasks to their default form; the practitioner rungs catch a chart that fights its task and justify a table over a chart on exact-value data; the senior rung composes the right form for a mixed brief and defends the choice.

Rung 1 — Match task to form (explorer)

Loading exercise…

Rung 2 — Catch the mismatched chart

Loading exercise…

Rung 3 — Table or chart?

Loading exercise…

Rung 4 — Compose the right form (senior)

Loading exercise…

Try it yourself

A scratch console for this page's ideas — ungraded, nothing you run here is recorded.

Scratch console

A scratch console with the scientific stack (pandas, numpy, scikit-learn). Runs on the server — no network, resource-limited and measured.

Output appears here.

Where next?

Continue

Scales, Axes & Legends →

Chart Craft · P14

Later in Chart Craft

Go up a level