dataviz/docs/06-condense.md

359 lines
15 KiB
Markdown
Raw Permalink Blame History

This file contains invisible Unicode characters

This file contains invisible Unicode characters that are indistinguishable to humans but may be processed differently by a computer. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

# CONDENSE Increase information density
CONDENSE covers all aspects of increasing information density in reports and
presentations.
_Increasing information density_ means that all reports and presentations
include all information that is necessary to understand the respective message
on one page.
This chapter covers using small components, utilizing space, as well as adding
data, elements, and objects.
1. [Use small components](#co-1-use-small-components)
2. [Maximize use of space](#co-2-maximize-use-of-space)
3. [Add data](#co-3-add-data)
4. [Add elements](#co-4-add-elements)
5. [Add objects](#co-5-add-objects)
## CO 1 Use small components
The need for a higher level of information density requires to display all
objects, elements, and signs as small as possible, while still being legible.
Different technical parameters apply to printed material, screen displays, and
projected slides.
## CO 1.1 Use small fonts
![Figure CO 1.1: Use small fonts](img/co-1.1.png)
In general, avoid oversize fonts. They needlessly waste space, see Figure CO
1.1.
## CO 1.2 Use small elements
![Figure CO 1.2: Use small elements](img/co-1.2.png)
Small elements increase clarity. Large-scale symbols and highlights are not more
suitable than smaller symbols and highlights, see Figure CO 1.2.
## CO 1.3 Use small objects
![Figure CO 1.3: Use small objects](img/co-1.3.png)
The size of charts and tables in reports and presentations should not be as
large as possible, rather as small as possible yet only so small so that the
objects and all its details and labels can be read easily. This provides room
for more information and therefore better understanding of the context,
see Figure CO 1.3.
## CO 2 Maximize use of space
Utilizing free space is the fastest and easiest way to increase information
density. Make better use of needlessly wide margins and frames, or blank or
little used pages by filling them with helpful data pertaining to the context.
## CO 2.1 Use narrow page margins
![Figure CO 2.1: Use narrow page margins](img/co-2.1.png)
The page layout is often dominated by corporate design standards not made for
high information density but for attractive design, sacrificing valuable space
to layout elements such as extra wide page margins, see Figure CO 2.1.
## CO 2.2 Reduce empty space
![Figure CO 2.2: Reduce empty space](img/co-2.2.png)
Reduce empty space to increase information density. This applies not only to the
page layout (see Figure CO 2.1) but also to the layout of report objects such as
charts and tables (see Figure CO 2.2).
## CO 3 Add data
Adding more data to an existing visualization increases information density and
helps better understand the context.
## CO 3.1 Add data points
![Figure CO 3.1: Add data points](img/co-3.1.png)
Displaying more data points does not jeopardize the comprehension of numerical
data. For example, a monthly statistic of staff numbers over twelve months in a
year would be understood just as quickly as for the same data series with twelve
months for each of the last three years in other words, a total of 36 data
points instead of twelve. Usually, interesting relationships are only detected
with an increased number of elements in a data series (see Figure CO 3.1).
## CO 3.2 Add dimensions
![Figure CO 3.2: Add dimensions](img/co-3.2.png)
A very useful way to increase information density is to show more than two
dimensions of a business situation. A chart with only one dimension (such as in
a pie chart), visualizes only mundane things easily stated in a simple sentence.
Already charts with two dimensions can yield very interesting relationships
yet those charts with three and more dimensions yield structures leading to
completely new insights (see Figure CO 3.2).
## CO 4 Add elements
It is often appropriate to use two or more basic chart types (either horizontal
or vertical) to build _combined charts_ with a higher information density.
_Combined charts_ are treated as one entity as opposed to multiple charts.
_Combined charts_ can be built both out from horizontal or vertical charts.
There are three types of combined charts depending on their type of combination:
_Overlay charts_, _multi-tier charts_, and _extended charts_. Additionally,
chart elements can be embedded in tables and explanations can be integrated.
## CO 4.1 Show overlay charts
![Figure CO 4.1: Show overlay charts](img/co-4.1.png)
In an _overlay chart_, two or more basic charts overlap. These overlapping
charts always use the same category axis.
_Overlay charts_ can facilitate comprehension such as in the combination of the
development of sales (a series of columns) and the return on sales in percent (a
line). However, this approach can only be used for a few chart combinations,
see Figure CO 4.1.
![Figure CO 4.1-1: Overlay chart with lines and columns using different value
axes](img/co-4.1-1.png)
_Overlay charts_ frequently use different value axes. A _column chart_
representing a measure (e.g. sales) combined with a _line chart_ representing
another measure (e.g. employees) is a typical example.
![Figure CO 4.1-2: Overlay chart with columns and lines using the same value
axis](img/co-4.1-2.png)
Sometimes, the same value axis is used as well. A _column chart_ representing a
measure (e.g. sales) combined with a _line chart_ representing the same measure
(e.g. industry average) is a typical example for such an _overlay chart_.
![Figure CO 4.1-3: Overlay column chart with integrated
variances](img/co-4.1-3.png)
Column or bar charts with _integrated variances_ (variances displayed within the
columns or bars) are other typical example for _overlay charts_ using the same
value axis (see the last two figures).
Compared to two-tier charts, this presentation of two data series uses much less
space. The disadvantages, though, are twofold: First, it is difficult to label
the data of both the primary and secondary chart. Second, the development over
time (horizontal axis) respectively the structure (vertical axis) of the primary
chart is difficult to see.
![Figure CO 4.1-4: Overlay bar chart with integrated
variances](img/co-4.1-4.png)
Suggestion: If there is enough space, use multi-tier charts instead.
## CO 4.2 Show multi-tier charts
![Figure CO 4.2: Show multi-tier charts](img/co-4.2.png)
_Use multi-tier charts_ to increase information density by adding additional
tiers to the same category axis for analyses on the same basic data. Multi-tier
charts are most frequently used for displaying variances along with the basic
values, see Figure CO 4.2.
![Figure CO 4.2-1: Horizontal multi-tier charts](img/co-4.2-1.png)
In a _two-tier chart_, a _secondary chart_ is shifted in parallel to the
category axis of the _primary chart_. For horizontal charts the secondary chart
appears above the primary chart, for vertical charts the secondary chart appears
to the right of the primary chart.
In both cases, the _category axes_ of the primary charts are reduplicated in the
secondary charts, usually having a different semantic scenario design.
![Figure CO 4.2-2: Vertical multi-tier chart](img/co-4.2-2.png)
Both the primary and the secondary charts have their own value axes. Value axes
showing the same currency or the same physical unit should be scaled
identically.
In a _three-tier chart_ a third chart appears above a horizontal or to the right
of a vertical two-tier chart. In special cases, more than three tiers can be
combined.
Improve the interpretation of a primary chart showing grouped bars for actual
and plan data by adding variances. In the second and third figure a secondary
chart with absolute variances and a tertiary pin chart with relative variances
are combined.
## CO 4.3 Show extended charts
![Figure CO 4.3: Show extended charts](img/co-4.3.png)
An _extended chart_, arranges additional charts next to the primary chart by
virtually extending the category axis. This way of increasing information
density often is used when displaying context information such as market
averages or competitor figures, see Figure CO 4.3.
![Figure CO 4.3-1: Horizontal extended chart](img/co-4.3-1.png)
For horizontal charts, additional charts appear to the left or right of the
primary chart, for vertical charts, above or below. In both cases, position the
_category axes_ of the additional charts on a virtual extension of the category
axes of the primary chart.
In an extended chart, use the same value axis for both the primary and the
additional charts.
Improve the interpretation of a primary chart by adding extended charts showing
the same values from a different perspective. In the figure on the left, a
secondary _grouped column chart_ at the right hand side shows the monthly
average.
## CO 4.4 Embed chart elements in tables
![Figure CO 4.4: Embed chart elements in tables](img/co-4.4.png)
Increase the information density of tables by using _chart elements_, see Figure
CO 4.4. Bars, warning dots, sparklines, and traffic lights are the predominant
chart element types in tables.
**Table bars**
_Table bars_ are bar charts integrated into tables. The categories of these bar
charts must correspond to the rows of a table. Both single bar charts with
single bars or pins and waterfall bar charts are powerful means to visualize the
absolute figures and variances in tables. Most recommendations concerning
vertical chart types can be applied to _table bars_.
**Warning dots**
Not to be confused with _traffic lights_, _warning dots_ can be a good solution in
highlighting important negative, positive, or questionable parts of a table. It
is important to use only very few warning dots in one table.
**Sparklines**
Omit _sparklines_ if not scaled properly. Individually scaled sparklines can be
misleading because small fluctuations in a series of other small fluctuations
look the same as big fluctuations in a series of big fluctuations. However,
sparklines with proper scaling (e.g. indexed) can be helpful.
**Traffic lights**
_Traffic lights_ contain little information, as they represent no more than
three (red, green, yellow) states. Use them only if there is no more information
to be conveyed than those two or three states (e.g. “yes” or “no”). In all other
cases, replace traffic lights with more suitable means of representation, such
as _table bars_.
## CO 4.5 Embed explanations
![Figure CO 4.5: Embed explanations](img/co-4.5.png)
Both the density of information and the level of comprehension increase when
explanations are embedded into charts and tables (this applies to written
reports and handouts only). When the explanation refers directly to the visual
presentation in question, it helps to establish a connection and speeds up
comprehension, see Figure CO 4.5.
## CO 5 Add objects
Reports and presentation material consist of one or more _pages_. The content of
one page can be viewed together without referring to other content, e.g.
flipping to other pages.
Reports and presentation material often arrange more than one chart on one page.
While this increases information density and fosters a higher level of
comparability, it presents a design challenge: A uniform notation concept and
consistent scaling are even more important than on pages with single charts.
The most important types of pages with multiple objects are small multiples and
multi-charts (including _ratio trees_).
## CO 5.1 Show small multiples
![Figure CO 5.1: Show small multiples](img/co-5.1.png)
Substantially improve the comprehension of complex relationships by displaying
charts of the same type and the same scale on the same page. These charts are
called _small multiples_, see Figure CO 5.1.
Typical applications are charts with different countries, products, or projects
placed next to each other. Of course, there is an upper limit to the number of
charts on one page, depending mainly on the page- and font-size used.
![Figure CO 5.1-1: Screen page with small multiples](img/co-5.1-1.png)
Showing _small multiples_ is a good way to compare a set of up to around 25
charts. Instead of exceeding this number on one page, a new chart called
“Others” containing the accumulation of all other elements could be a solution.
As mentioned in the chapter “CHECK Ensure visual integrity”, all small
multiples must use the identical scale.
Working with _small multiples_ can be difficult if certain charts show
significantly bigger values than others. Using a different scale for a chart
with bigger values is not a feasible option, increase the size of this chart
instead.
## CO 5.2 Show related charts on one page
![Figure CO 5.2: Show related charts on one page](img/co-5.2.png)
Different from small multiples, _related charts cover different topics
(different measures) on one page._ They mostly use different scales, too. This
arrangement of charts on one page is sometimes called _multi-charts_. But the
term _multi-charts_ fails to underline the fact that these charts must have a
useful relationship. It does not make sense to arrange several, completely
unrelated charts on one page.
This approach offers high data density and a higher level of comparability but
it can be a demanding visual and technical challenge as a uniform notation
concept, clear terms, and an understandable scaling prove even more important
(see Figure CO 5.2).
![Figure CO 5.2-1: Page showing a ratio tree](img/co-5.2-1.png)
Consistent scaling of _multi-charts_ can be difficult. Sometimes different
scales for the same unit or measure are inevitable. In this case, clearly
indicate the use of a different scale by an appropriate mean, e.g. scaling
indicators.
_Ratio trees_ are multi-charts showing root causes. Use ratio trees to prove or
explain a specific issue. Pointing out the assumptions and root causes of
variances or temporal evolvements improves understanding and is more convincing,
too. In general, the _ratio_ is broken down into its components (mostly from
left to right). Thus individual charts, preferably identical size, are arranged
in a tree shape structure.
Consistent scaling of _ratio trees_ can be difficult. Sometimes different scales
for the same unit or measure are inevitable. In this case, clearly indicate the
use of a different scale by an appropriate mean, e.g. scaling indicators.
A typical example of a page showing a _ratio tree_ is the “Return on asset”
tree.
## CO 5.3 Show chart-table combinations
Combining charts and tables on a page is not to be confused with the integration
of chart elements in tables.
_Chart-table combinations_ cover situations where a separate chart is added to a
page with a table or vice versa. In general, such a combination is very useful
if both objects display supplementary data. Tables simply listing the numbers of
a chart are superfluous in most cases (see also UN 2.3 “[Unify the position of
legends and labels](09-unify.md#un-23-unify-the-position-of-legends-and-labels”).
## CO 5.4 Show charts and tables in text pages
Embedding illuminating charts and tables in the text of a written report helps
the reader understanding the message.
Always position charts and tables in close proximity to the phrase carrying the
message, which the chart or table supports.
Text pages should contain a title element like other pages. Also use a title
and, if possible, a message for every chart and table embedded in a text page.
[← Avoid Clutter](05-simplify.md) | [Ensure visual integrity →](07-check.md)