Abstract
This chapter lays the groundwork for the chapters to come. In it is described the basics of how a visualization is constructed, giving a rough overview of the data pipeline and describing the components of a general visualization. I define a visualization in terms of a set of orthogonal features that can be composed to form a complete visualization. These “building blocks” include the coordinates system; the type of element (line, point, bar); color, shape, and other mappings from data to appearance; statistics; and paneling. These definitions allow us to talk about visualizations sensibly and precisely, as well as to classify visualizations by their features. In later chapters we will see how different choices of features serve different graphical goals, and we will then be able to compose the features we want to serve our goals, resulting in an effective visualization.
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsNotes
- 1.
Results from this survey have been published in a number articles and several books, of which the reference cited above is only one of many interesting articles.
- 2.
One reason not to ban such seeming nonsense is that you never know how language is going to change to make something meaningful. A chart that a designer might see no use for today becomes valuable in a unique situation, or for some particular data. “The tasty age whistles a pink” might be meaningless, but “the sweet young thing sings the blues” is a useful statement.
- 3.
In Sect. 2.3 we will explain a little more about statistics. In particular we will deal with the use of weight variables, the use of which is the best way to describe this data set for most purposes.
- 4.
The details of drawing a boxplot are technically quite tricky, especially in the presence of weighted data, for which Tukey does not provide much help. For large amounts of unweighted data these details may not be apparent, but for small data sets and for weighted data sets it is possible to get different displays from different graphical packages. However, since the boxplot was designed primarily for exploring data, these minor technical differences should not affect the overall goal of discovering patterns and trends.
- 5.
In this context, the terms smooth and predictor are being used interchangeably. To be more formal, we would say that a smoothing statistic can be used as a predictor, rather than using the two terms interchangeably.
- 6.
The formula for an Epanechnikov kernel is defined as 0 outside the window and
$$\frac{3} {4}{\Bigl (1 -{\bigl ({\frac{x} {h}\bigr )}}^{2}\Bigr )}$$when − h < x < h for a window width of h. This kernel minimizes the asymptotic mean integrated squared error and can therefore be thought of as optimal in a statistical sense.
- 7.
Home runs are big hits that in today are virtually always hit out of the field of play. Stolen bases, in contrast, are when the runner advances to the next base without making a hit at all. In a sense, they are opposite types of play that advance the bases for the batting team.
- 8.
These represent players in the American League, which has a position called the designated hitter, or DH. Under rules in the American League the pitcher, who is usually a weak hitter, can be replaced by a player whose only job is to hit the ball; he has no defensive role at all. Rather than being surprised that this makes a big difference in hitting statistics, we might instead be surprised that this difference does not appear until 1994 – the rule had been in effect since 1973.
- 9.
Figure 2.26 shows a limitation on the use of color to denote groups. With 50 different lines, it is hard to perceive differences between all pairs of colors clearly. In this figure, the fact that each line forms its own group is the strong grouping construct – the color is added only to help the viewer separate the lines when they cross or intersect each other.
- 10.
A 100% opaque object is 0% transparent and vice versa. Using opacity is preferable as the more opaque something is, the more visually prominent, and so opacity can usefully be used as an aesthetic when mapped to counts, weights, sums, and the like, since larger values of these statistics should be shown more prominently.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
Copyright information
© 2010 Springer New York
About this chapter
Cite this chapter
Wills, G. (2010). Framework. In: Visualizing Time. Statistics and Computing. Springer, New York, NY. https://doi.org/10.1007/978-0-387-77907-2_2
Download citation
DOI: https://doi.org/10.1007/978-0-387-77907-2_2
Published:
Publisher Name: Springer, New York, NY
Print ISBN: 978-0-387-77906-5
Online ISBN: 978-0-387-77907-2
eBook Packages: Mathematics and StatisticsMathematics and Statistics (R0)