You are viewing a public preview
Sign in to save your progress, do assignments, join discussions and get feedback.
Purpose of commonly used charts; comparison, trend, composition, relationship and geographic charts; matching chart types to data and analytical questions. The slides cover tables, heatmaps, line charts, pie charts, bar charts, histograms, scatter plots, bubble charts and treemaps for hierarchical data, followed by a chart identification exercise.
Learning outcomes
The lecture slides for this lesson are below.
By the end of this lesson you can:
Key idea: Start from the question, not the chart. Decide what you want to show (comparison, trend, composition, relationship, distribution or location) and what data you have, and the chart type usually follows.
The chart chooser in the slides starts with one question: "What would you like to show?" It then branches into four families, to which business work adds geography:
| Objective | Typical question | Common charts |
|---|---|---|
| Comparison | Which item is bigger? How did items change over time? | Bar or column chart; line chart for many periods |
| Composition | What share does each part contribute? | Pie, stacked bar or column, 100% stacked, treemap, waterfall |
| Distribution | How are values spread out? | Histogram; scatter plot for two variables |
| Relationship | Do two or more measures move together? | Scatter plot; bubble chart for three variables |
| Geographic | Where does it happen? | Symbol map, filled map |
A table is the simplest visualization: items in rows, attributes in columns. It is very effective for precise values and widely accepted, but it shows trends poorly, rarely conveys one clear message and becomes overwhelming with a lot of data.
A heatmap needs two categorical attributes (rows and columns) and one quantitative attribute shown as colour. It is good for finding clusters and outliers and scales to very many items, but people can only tell apart around ten levels of colour, so it shows patterns rather than exact values.
A treemap shows hierarchical data as nested rectangles, with a quantitative attribute at the leaf level encoded as rectangle size. It scales to very many leaf nodes and shows both the parts and the hierarchy, for example Category containing Sub-Category. Comparing rectangles of similar size is harder than comparing bars.
A sales manager at the course's office-supplies retailer brings four new questions. Reason through each one.
In each case, the decision started from the objective (comparison, trend, relationship, composition) and the number and type of fields.
Flashcards · Common Chart Types in Data Visualization
9 wordsA grid of two categorical attributes in which colour encodes a quantitative value, used to spot clusters and outliers.
A chart that connects points over a continuous axis, best for showing trends over time.
A chart using bar length to compare values across categories; axes must start at zero.
A circle divided into slices showing parts of one whole; positive values and few slices only.
A chart of how often numeric values fall into bins; reveals shape, skew and outliers.
A chart of two numeric variables as points, used to reveal correlation, clusters and outliers.
A scatter plot whose marks are sized by a third measure.
Nested rectangles sized by a measure, showing hierarchical part-to-whole data.
A guide that selects charts by objective: comparison, composition, distribution or relationship.
No software needed. For each business question, write down (a) the chart type you would use and (b) one sentence explaining why. Then compare with the suggested answers below.
Suggested answers: 1 line chart (trend over continuous time) · 2 sorted horizontal bar chart (comparison across many categories; a pie would be unreadable with 13 slices) · 3 pie or 100% stacked bar (part-to-whole with few parts) · 4 scatter plot (relationship between two measures) · 5 filled or symbol map, or a sorted bar chart if exact comparison matters more than location · 6 histogram (distribution of one measure).
In Module 3 you will build these charts in Tableau with the course dataset.