Data Visualization Midterm
What is unique about a key attribute?
It can be used to look up items.
An adjacency diagram has what advantage over node-link diagrams?
It can use a length encoding.
Shadows can be used to enhance depth cues. Which is not a problem with using shadows that is suggested by the book?
It requires the user to build a 3D model.
A Dorling chart is specific from its general type of chart in what way?
It uses circles.
Which is not a difference between sets and lists?
Items in sets have restricted attribute types.
The order of layers in streamgraphs
depends on what you're trying to show.
Which is not a typical synonym for attribute in visualization?
description
The question of how many bins are available for use within a perceptual channel is a question of
discriminability
In a bubble tree layout you can see a node's depth by looking at its
distance to its parent
Link marks, containment, usage of identity channels, and proximity are all approaches to
grouping
A good way to lower the load on memory when zooming is to
have an overview window.
Force-directed placement of node-link nodes is unlikely to yield good results when
the numbers of nodes is roughly four times greater than the number of links.
Using different encodings for the same data in coordinated juxtaposed views is called
multiform design.
The difference between multiform and small multiples is
multiform uses the same data with different encodings and small multiples uses different data with the same encodings.
Connection and containment marks are used for what type of datasets?
network
A weakness of node-link layout algorithms is that
node proximity can mislead the user.
The most common visual encoding idiom for tree and network data is
node-link
A dendogram is an example of what type of visualization?
node-link diagram
If you're interested in local topological structure, such as small numbers of hops from nodes, you should probably use a
node-link diagram
The most powerful depth cue is
occlusion
Recursive subdivision partitions items
on multiple attributes.
How many points along the ray of our line of sight can we see?
one
Bar charts encode
one quantitative and one categorical attribute
Animated transitions are most effective when
only a few things change.
Which is not a typical channel used in vis?
opacity
Line charts should only be used when the key is
ordered
Change blindness means we miss even big changes when
our attention is directed elsewhere.
How are axes in parallel coordinates represented?
parallel instead of orthogonal to each other
What is made especially powerful by the privileged status of spatial position as the highest ranked channel?
sorting
In LiveRAC, line charts are squished down to
sparklines.
What type of channels does arrange cover?
spatial
What visualization element leads to the most accurate decoding of numerical data?
spatial position
Scientific visualization is characterized by
spatial positioning being given with the data.
Which performs the best for point estimation tasks?
2D point clouds
Force directed layout is used for what data structure?
graph
Height is (1) while elevation is (2).
(1) sequential (2) diverging
How do we surpass the limitations of our own internal cognition and memory?
By using external representations.
Place the channels in order of effectiveness. (Only a subset of channels is shown here.) - position on common scale - Volume (3D Size) - Length - Position on unaligned scale - Angle
- position on common scale - Position on unaligned scale - Length - Angle - Volume (3D Size)
Place the channels in order of effectiveness.
1 - Spatial Region 2 - Color Hue 3 - Motion 4 - Shape
A mark that is a line is
1D
How many attributes are used to partition the data in the grouped bar chart of figure 12.8(a)?
2
How many keys does a typical stacked bar chart use?
2
Points visualized in 2D can encode how many quantitative attributes using only spatial position channels?
2
According to Colin Ware, how many dimensions do we see in?
2.05D
Which operators apply to ordinal attributes? (There may be more than one.)
< AND =
Which operators apply to categorical attributes? (There may be more than one.)
=
Which operators apply to quantitative attributes? (There may be more than one.)
= AND <
Suppose I have items with three quantitative attributes listed in order of greatest to least importance: A, B and C. Match the attributes with channels that I should use. Not all the channels listed will be used. - Color hue - position on common scale - color saturation - spatial region - length
A - position on common scale B - color saturation C - length
Why are most visualizations ineffective?
A possible design is a poor match with the properties of the human perceptual and cognitive systems.
What is the difference (according to the book author) between angle and tilt?
Angle is judged with respect to another line; tilt is judged against the display frame.
Hops are
Discrete
Match the terms relating to sharing data among views.
Each view shows all the data -- Shared data A view shows a subset of the other view -- Overview-detail Views show different partitions of the data -- small multiples
Treemaps are examples of what type of chart?
Enclosure diagram
How are graduated symbol maps different from cartograms?
Graduated symbol maps place symbols over an underlying map.
Brushing-and-linking is a technique used frequently in what charts?
SPLOM
Scatterplot matrices are also known as
SPLOMs
The full adjacency matrix need not be shown when
The network is undirected
Why is there so much more cognitive load in relying on navigation for understanding 3D structure?
The user must use internal memory to remember previous viewpoints.
Why is scalability an important concern?
Vis systems are almost always used for larger datasets than they were designed for.
Anscombe's Quartet is a dataset that is used when arguing what?
We should show data in detail when possible.
What is a reason that makes multivariate data (i.e. data in high dimensions) hard to represent?
We're used to picturing things in 2D and 3D.
When is 3D absolutely justified?
When visualizing structures that are inherently 3D
A pie chart can be represented by
a bar in a stacked bar chart.
A network is
a dataset type.
Something that is generally lost when moving from a standard 2D display to virtual reality is
a familiar workspace.
A simple bar chart splits bars into regions based on
a key.
The basic graphical image in a vis is called
a mark.
In biology, a gene as one key and time as another would be represented as
a multidimensional table.
What is the definition of a link data type?
a relationship between items
Bar charts are usually ordered. What defines the ordering?
a value attribute
What is a cell in a table?
a value representing a given attribute for a given item
A scatterplot is an example of
a vis idiom.
Heer and Bostock used crowdsourcing to measure
accuracy
A derived table dataset with two key attributes is used in
adjacency matrix views
Blink comparators
allow jumping between two frames.
Which of the following are examples of time-varying semantics?
animal locations measured at regular time intervals
Polar area charts
are an improvement of pie charts.
Selection and highlighting
are closely tied.
Constrained navigation often involves the visualization system computing a smooth camera trajectory to a viewpoint that is
better for viewing selected objects.
Which elements of networks can have attributes?
both nodes and links
What is another name for linked highlighting?
brushing
What is a name for scatterplots using size to portray an additional quantitative attribute?
bubble plots
Slicing
can be used to eliminate multiple dimensions at once.
All interactive idioms involve a view that
changes over time.
What is a general definition of navigation?
changing the viewpoint
What controls the appearance of marks?
channels
Using raw data values rather than normalized values is a common error in what kind of chart?
choropleth map
Heatmaps use what channel for the value attribute?
color
Whether all the data is provided at once is a question of
data availability.
What, why, and how are answered with what trio?
data, task, idiom
Which hierarchical representation places all leaf nodes of the tree at the same level?
dendrogram
Which is most closely related to a line chart?
dot chart
Given enough display space, how many discriminable bins are there for shape?
dozens
Parallel coordinates scales to
dozens of attributes.
What principle dictates that the importance of the attribute should match the salience (noticeability) of the channel?
effectiveness principle
Using spatial position for categorical attributes would violate what principle?
expressiveness
What principle dictates that the visual encoding should express all of, and only, the information in the dataset attributes?
expressiveness principle
Interpolation is meaningful in the context of what dataset type?
fields
The outcome of navigation is a combination of
filtering and aggregation.
Reducing the amount of data that is shown generally involves
filtering.
Charles Minard's depiction of Napoleon's march on Moscow is what type of chart?
flow map
Which is the form of navigation most familiar to us in the natural world?
geometric zoom
What alternative to the choropleth map can represent more dimensions (as specifically pointed out in the paper)?
graduated symbol map
Space-filling layouts strive for
high information density.
Parallel coordinates are limited due to
high learning curve for users.
Motion channels are most appropriate for what?
highlighting
When you have only very little space for an area chart, what alternative has been found to be more effective?
horizon graph
Visual popout is possible because
human perception can be massively parallel.
What type of channel is hue?
identity
What type of channel is shape?
identity
What are the two main classes of channels?
identity and magnitude
As the size of a horizon graph decreases, the number of bands
increases
Which hierarchical representation is used ubiquitously in operating systems to represent a file system?
indented tree
Choose the best answer. Which line chart shows percentage changes for a collection of time-series data based on a selected point in time?
index chart
Which is not a datatype?
integer
Horizontal size and vertical size are two channels that are
integral
A great solution to handling complex data in visualizations is what?
interactivity
A stacked time-series area graph, or simply, stacked graph, has interpretation issues that can be helped by using what?
interactivity
According to the textbook, what unleashes the full power of linked views?
interactivity
The curvature channel
is not generally a good choice.
Accuracy of our perception of angle
is not uniform.
A weakness of the motion channel is that
it draws attention away from other channels.
An alternative to animated transition is the
jump cut.
Which scales better in the number of views?
juxtapose
The difference between juxtapose and changing view is
juxtapose uses space and changing view uses time.
Regions using superimpose are broken into
layers.
What are the unique values of categorical values called?
levels
Banking to 45 degrees is used in
line charts.
Seeing only one point along the ray of the depth axis is called
line-of-sight ambiguity.
The number of links divided by the number of nodes is called
link density
What data type do tables generally not include?
links
Maureen White suggests using which color channel for the most important attribute?
luminance
What type of channel is length?
magnitude
What type of channel is luminance?
magnitude
Map projections deal with
mapping the surface of the earth to 2D.
Which is not a term for a region with some visual encoding?
mask
An alternative to a force-directed layout of a graph is what?
matrix
Which idiom can scale to one million edges?
matrix
What highlighting technique often outperformed more traditional techniques in a study?
motion coding
Small multiples
partitions the data among views.
The most fundamental expression of the expressiveness principle is that ordered data should be
perceived as ordered.
The psychophysical power law of Stevens measures
perceived sensation as a function of physical intensity.
Which of the following is pointed out in the book as being great for visual art but bad for data vis?
perspective
Something that is generally reduced when moving from a standard 2D display to virtual reality is
pixel resolution.
Faceting comprises both juxtapose and superimpose. Juxtapose
places views spread out in space.
Whether there are 5 or 50 items is irrelevant for channels that show good
popout
A dense layout
provides an overview.
When data is periodic
radial layout may be a good choice.
What color is Santa Claus Day shown in?
red
Given a dataset with n attributes, if the vis uses n+1 channels then one of the attributes is encoded
redundantly
Weber's Law relates to
relative vs. absolute judgments.
Fields are often obtained through
sampling from a continuous domain.
What is a common use for scatterplots?
seeing correlations
What is the term for the real-world meaning of a dataset?
semantics
Partitioning
separates items based on attributes.
Treemaps are often used when hierarchies are
shallow
Matrix reordering
shows large-scale patterns between attributes.
A grouped bar chart
shows partitioned data.
Good visual representations can replace cognitive calculations with what?
simple perceptual inferences
What links the same items together between table columns?
slope graphs
Which chart can simplify understanding of multiple time series, bar charts, pie charts, and maps?
small multiples charts
1D lists are
sometimes better than 2D layouts.
Which can be used with categorical attributes?
spatial region
In the context of data visualization, the verb facet means to
split
Matrix views are
stable
"Stream graph" is another name for what graph?
stacked graph
Hatching and cross-hatching are often synonyms for what?
stippling
Orientation, scale, and contrast compose what visual channel?
texture
Selecting by levels selects all items
that share a value for an attribute.
Animated transition is most useful when
the amount of change is limited.
An index chart bases time-series data on a selected point in time. What is that point called?
the index point
In semantic zoom
the number of available pixels determines the representation.
Parallel coordinates can be better than scatterplots when
the number of dimensions exceeds 2.
Which of the following do you think is not a reason for the lack of numeric values on the y-axis of a stacked graph?
the spatial distribution of values would be highly non-uniform
Multiform views are a good design choice when
there are too many attributes to show without clutter.
Why might lines in a vis have different widths from each other? Choose the best answer.
to encode an additional quantitative attribute
What deals with how nodes in a general network are connected?
topology
Which is one of the best-justified uses of animation?
transition
A dataset type that is a specific type of network is a
tree
Containment is used in
treemaps
Main effects ordering allows users to see
trends and outliers.
Matrix alignment has
two keys.
A scatterplot encodes
two quantitative attributes
What does a scatterplot show?
two values
A weakness of matrix views is that they're
unfamiliar to many people
Deciding whether your vis has met your design goals is called what?
validation
A key is important to find what?
value attributes
In which dimension do we perceive differences as being most important?
vertical
Stereoscopic depth cues are most useful for objects that are
within arm's reach.
Short-term memory is called
working memory.
The main limitation to layers is that
you can't use very many in the same view.
