Introduction
“A picture is worth a thousand words.” Within the realm of science
and social science, a picture, whether in the form of a graph or plot,
can be the difference between confusing a reader and being able to
provide your reader with a concise picture of your data, model and/or
outcome.
Even the most germane graph or plot is, by convention, divorced
from the text in which is embedded. While the author or authors of an
article, thesis or dissertation can attempt to frame visual
representation with the most appropriate possible text, this can
sometimes disrupt the flow of discussion and can create space problems
for journal submissions. Do you exclude the graph or plot if you’re
operating under space constraints? Do you attempt to re-arrange an
entire chapter of a thesis or dissertation in order to place a graph
or plot in a more appropriate location?
For some categories of data, the solution lies with a tool called
sparklines. Developed from concept through implementation by Dr.
Edward Tufte (http://www.edwardtufte.com/tufte/
), Professor Emeritus at Yale University, sparklines allow for the
in-line representation of data in a document.
What Are Sparklines?
Tufte defines sparklines as “small, high-resolution graphics
embedded in a context of words, numbers and images. Sparklines are
data-intense, design-simple, word-sized graphics.” Sparklines allow
the researcher to display data in-line and, most importantly, in
context.
Sparklines have different formats to match different types of data
(from
http://www.sparklines.org ): Series data on the Federal deficit
from the Office of Management and Budget from 1983-2003 (
),
the five-year daily closing stock price of Cisco (
)
or the postseason performance of the 2004 Boston Red Sox en route to
their first World Series championship since 1918 (
).
Implementing Sparklines
The Bundesliga (http://www.bundesliga.de/
) is Germany’s top-level professional football league. The Bundesliga
season is 34 games long; while this seems like a small number of games
given the length of the season (more like our MLB season than our NFL
season), there are numerous other team and player commitments that run
concurrently with the Bundesliga season: European competition for the
more successful clubs (Champions League and UEFA Cup) and
international competitions which halt domestic play since many players
are absent from their clubs representing their respective nations.
As any fan of sports can attest, there are an overwhelming amount
of numbers associated with your favorite player, your favorite team
and the league in which your team plays. For example, if you wanted to
know the record of every team in the Bundesliga from a particular
season, you could simply represent it in a table like the following
one (modified from
http://www.bissantz.de/sparklines/demos/bundesliga/season_2004-2005.html
):
P
Team
W D
L Goals
GD TP
1.Bayern München
24 5
5 75:33
42 77
2.FC Schalke 04
20 3
11 56:46
10 63
3.Werder Bremen
18 5
11 68:37
31 59
4.Hertha BSC Berlin
15 13
6 59:31
28 58
5.VfB Stuttgart
17 7
10 54:40
14 58
6.Bayer Leverkusen
16 9
9 65:44
21 57
7.Borussia Dortmund
15 10
9 47:44
3 55
8.Hamburger SV
16 3
15 55:50
5 51
9.VfL Wolfsburg
15 3
16 49:51
-2 48
10.Hannover 96
13 6
15 34:36
-2 45
11.SV Mainz 05
12 7
15 50:55
-5 43
P=Place finished,
Team=Team Name, W=Win, D=Draw, L=Loss, Goals (for:against), GD=Goal
Differential, TP=Total Points
While this table provides us with the basic information we need to
see how all the teams in the Bundesliga performed in the 2004-2005
season, wouldn’t it be nice to be able to display a team’s game by
game performance in terms of wins and losses along with this
information?
Thinking along conventional lines, displaying game by game
performance in a tabular form would take up a ridiculous amount of
room (34 games x 18 teams = 612 total games); in fact, so much room
would be taken up that you will rarely find such a display outside of
a sports almanac or season-in-review publication.
Implementing the same idea graphically using sparklines is possible
in the space of a table. Note the in-line graphics under the column
labeled “Course”: A blue uptick is a win and a red downtick is a loss.
A dash represents a draw. In a comparable amount of space to what it
takes to present the team names, 612 data points of wins, losses and
draws can be displayed in a meaningful graphic:
P Team
W D L Goals GD TP
Course
1.Bayern München
24 5 5 75:33 42 77 
2.FC Schalke 04
20 3 11 56:46 10 63 
3.Werder Bremen
18 5 11 68:37 31 59 
4.Hertha BSC Berlin
15 13 6 59:31 28 58 
5.VfB Stuttgart
17 7 10 54:40 14
58 
6.Bayer Leverkusen
16 9 9 65:44 21 57 
7.Borussia Dortmund
15 10 9 47:44 3 55 
8.Hamburger SV
16 3 15 55:50 5 51 
9.VfL Wolfsburg
15 3 16 49:51 -2 48 
10.Hannover
96 13 6 15 34:36 -2
45 
11.SV Mainz 05
12 7 15 50:55 -5
43 
12.FC Kaiserslautern
12 6 16 43:52 -9 42 
In Part 2 of this article coming in the September 2005 edition of
Benchmarks Online, I will cover the technical details of
implementing sparklines in Microsoft Word, Microsoft Excel, LaTeX,
HTML, PHP and Ruby. Below is a list of some links to information on
Sparklines, including discussions on Edward Tufte’s website and
Sparklines.org.
Until next month, happy computing!
Links:
