IS6335 Chap.5 Visualizing and Arranging Table Data
Visualizing and Arranging Table Data
The longest idiom list in the course
Module two opens on the most common dataset type. The published week four topics are the table data structure, mark and channel usage, and design idioms including the scatterplot, bar chart, pie chart, line graph and histogram, with a Tableau demonstration on table arrangement and an R demonstration on table visualization.
The list is easy to memorise; the selection is what carries marks.
Two questions settle most cases. How many attributes does the reader's question involve, and what type is each of them. A question about one quantitative attribute across categories is a bar chart.
A question about the shape of one quantitative attribute on its own is a histogram, which is a different idiom despite looking similar, because its horizontal axis carries binned values rather than categories.
A question about whether two quantitative attributes move together is a scatterplot, and replacing one of them with an ordered attribute such as time turns it into a line graph.
The connecting line is a claim
That last conversion is not cosmetic. A line graph asserts continuity between its points: it tells the reader that intermediate values existed and lay along the line.
Drawn across categories with no order, or across time points separated by large gaps, the assertion is false and the honest idiom is points without the line.
This is the first place in the course where an idiom choice makes a factual claim that the data may not support, and it will not be the last.
The week runs its material twice, once in a drag-and-drop tool and once in code, and that is not repetition for its own sake. The same idiom built both ways exposes different parts of the decision.
The tool makes the arrangement quick and the defaults invisible; the code makes every default explicit and the arrangement slow. Doing both is the cheapest way to discover which choices your tool was making on your behalf.
Ask what an item is before anything else
Before an idiom is chosen the table has to be described in the vocabulary of the previous chapter.
The demonstration table in the lectures is a set of orders with columns for an order identifier, an order date, an order priority, a product container and a ship date, and the material labels a row an item, a column an attribute and their intersection a cell.
In that table an item is one order, and the sentence decides a great deal. A bar chart of priority counts items, so it answers how many orders.
A bar chart of a value column sums an attribute across items, which is a different question and a different chart.
When a table has been joined from two sources the item is often no longer what you assume, and a row may represent an order line rather than an order, which silently multiplies every total in the view.
Derive the column the question is really about
The derive instruction applies here more often than anywhere else in the course.
The orders table carries an order date and a ship date and neither answers a question about delay; their difference does, and it is a quantitative attribute that did not exist until you created it.
The published week four sequence puts mark and channel usage immediately after the table structure for exactly this reason, because the mapping can only be as good as the columns you hand it.
Two choices that look cosmetic and are not
Stacking or grouping a bar chart is a question about the reader rather than about the data.
Stacking gives every category a common baseline for its total and denies one to every segment above the first. Grouping gives every segment a common baseline and denies one to the total. If the question is about totals, stack; if it is about comparing the parts, group; and if it is genuinely about both, produce two charts, because one baseline cannot serve both readings.
Saying which reading you gave up is worth marks, and pretending a single chart does both is not.
The pie chart is on the published list, so it is not forbidden. It encodes on angle, which is decoded less accurately than length, and it costs a legend.
It earns its place in one situation: a small number of parts that genuinely sum to one meaningful whole, where the reader's question is about share rather than about rank or magnitude. Two pies side by side almost never work, because comparing angles across separate circles is the hardest comparison in this chapter.
What this chapter covers
- 01
The published idiom list and how to select from it
- 02
How many attributes, and of what type
- 03
Bar chart against histogram, and why they differ
- 04
The line graph as an assertion about continuity
- 05
Items, attributes and cells on a real orders table
- 06
Deriving the column the question actually needs
- 07
Stacked against grouped, and what each gives up
- 08
When a pie chart is defensible
Turn a vague request into an idiom choice
- 2State what one item is and type the attributes.
- 2Name the derived attribute and its definition.
- 2Write the task as an action and a target.
- 2Choose an idiom and reject the obvious one with a reason.
Key terms
- Histogram
- An idiom for the distribution of one quantitative attribute, whose horizontal axis carries binned values rather than categories. Its bin width is a parameter that changes the shape it reports.
- Scatterplot
- An idiom giving two quantitative attributes to the two position channels, so that a dependency between them can be judged directly. It is the default answer when the target is correlation.
- Stacked bar chart
- A bar chart whose segments sit on top of one another, so the whole bar carries a total. Only the lowest segment has a common baseline, which is the cost of the totals it makes easy.
- Order line
- One row of a purchase rather than one purchase. Mistaking it for the item is a common source of multiplied totals when a table has been joined from two sources.
Visualizing and Arranging Table Data FAQ
When is a pie chart acceptable?
When there are few parts, when they genuinely sum to one meaningful whole, and when the reader's question is about share rather than about rank or precise magnitude. It is on the published idiom list, so treating it as forbidden misreads the course. What it costs you is accuracy, because angle is decoded less well than length, and a legend, because slices are usually labelled away from the marks.
Two pies placed side by side are the case to avoid outright, since comparing angles across separate circles is harder than any other comparison in the week's material.
My histogram looks different every time I change the bins. Which one is right?
None of them is right in isolation, and that is the point worth reporting rather than hiding. Bin width is a parameter nobody chose deliberately when the tool picked a default, and a wide bin can merge two peaks into one hump while a narrow one can split a single peak into noise. Redraw at two or three widths.
If a feature survives all of them it is in the data; if it appears at one width only, it is an artefact of the parameter. When a colleague claims two distinct groups, the check is to split the data by the attribute they think separates the groups and see whether each subset has a single peak.
Assessment move
For each of the five named idioms, write down in one line the task it answers and the claim it makes that the data has to support. The line graph's claim about continuity and the stacked bar's baseline cost are the two that get people into trouble.
Then, when you build any chart for a lab exercise, state the item first in writing; it takes one sentence and it catches the joined-table multiplication before you have built anything on top of it.
Working through Visualizing and Arranging Table Data in IS6335? Sia is AskSia’s AI Information Technology tutor — ask any IS6335 Visualizing and Arranging Table Data question and get a clear, step-by-step explanation grounded in how IS6335 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.
