Designing Better Graphs
Transcript of Designing Better Graphs
Designing Better Graphs
Matthew Wettergreen, PhD
Components of a Good Graph
● Proper data organization (MOST IMPORTANT)● Formatted without default settings● Designed for ease of readability for others
● Correct use of “preattentive stimulus”● Visual Hierarchy● Readable text● Properly formatted axes● Properly weighted lines● Appropriate color (only when necessary)
Identifying the Proper Format for Your Data:What Relationships are you trying to Show?
Relationship Example Question
RankIn what order are our proposed device designs based on the summed values from the Pugh matrix from high to low?
Part-to-whole
What portion of the market does our company command, and how does that compare to our competitors?
Time-series Is the traffic on our website increasing or decreasing?
DeviationTo what degree do our device’s readings vary relative to the expected numbers?
Distribution
What is the range of our employees’ salaries, and how many employees fall into each subset of that range in increments of $10,000?
Correlation
Is there a relationship between the stiffness of our construction material on the strength of our constructed shape?
Do Not Use Program Defaults!
● Thin lines
● Too small of text
● Box around data
● Often lacks labels
● Grey background
● Gridlines
● Bad colors
● Boxes around everything
Pre-attentive Processing
extremely fast, pre-conscious visual processing
How many 5’s are in this figure?
385720939823728196837293827
382912358383492730122894839
909020102032893759273091428
938309762965817431869241024
How many 5’s are in this figure?
385720939823728196837293827
382912358383492730122894839
909020102032893759273091428
938309762965817431869241024
Preattentive Processing
• There are some basic visual properties that can be detected immediately by low-level visual system
• These are identified from a “Pop-out” instead of a “Serial Search”
• Preattentive Processing relates to tasks that can be performed in less than 200 to 250 milliseconds on a complex display
Color (hue) is preattentive
Detection of red circle in group of blue circles is preattentive
Form (curvature) is preattentive
Curved form “pops out” of display
Don't Over Do It!Conjunction of Attributes
Conjunction targets generally cannot be detected preattentively
The red circle hides in the sea of red square and blue circle
Other Preattentive Examples
Which Preattentive Stimulus is Best for Quantitative/Qualitative Data?
Type Attribute Quantitatively Perceived?
Form Line Length Yes
Line width Yes, limited
Orientation No
Size Yes, limited
Shape No
Curvature No
Added Marks No
Enclosure No
Color Hue No
Intensity Yes, limited
Position 2-D Position Yes
Preattentive Stimuli has a Visual Heirarchy
Which preattentive stimulus is stronger?
Visual Hierarchy
Highlighting Trends Using Sorting
• Sorting and rearranging data lets users visually draw meaningful comparisons from the data
• Sort based on the relationship and pattern you are trying to flush out in this graph, not by convention (ie alphabetically)
Before After
Visual Hierarchy for Text
Amount of Sales by Salesperson
0
500
1000
1500
2000
2500
Dodsworth Suyama Davolio Fuller Peacock Leverling Arcoll Callahan Dusty
Salesperson
Am
ou
nt
of
Sal
es (
$)
16.5
13
13
108
Graph Text Must Have a Consistent Visual Hierarchy
The visual hierarchy of this graph is inconsistent
Visual Hierarchy of Text
Actual
Amount of Sales by Salesperson
Amount of Sales ($)
2500, 2000, 1500, 1000, 500, 0
Dodsworth, Suyama, Davolio, Fuller, Peacock, Leverling, Arcoll, Callahan, Dusty
Salesperson
Correct
Title (1st)
X-Axis Title (2nd)
Y-Axis Title (2nd)
X-Axis Data (3rd)
Y-Axis Data (3rd)
Data Labels (4th)
Choosing a Readable Font
Fine Legibility Poor Legibility
Serif San-Serif Serif San-Serif
Times New Roman Arial Script Gill Sans ITC
Palatino Verdana Broadway Papyrus
Courier Tahoma Old English Tempus Sans ITC
Amount of Sales by Salesperson
0
500
1000
1500
2000
2500
Dodsworth Suyama Davolio Fuller Peacock Leverling Arcoll Callahan Dusty
Salesperson
Am
ou
nt o
f Sa
les
($)
The visual hierarchy of this graph has been fixedand is now consistent
Properly Formatting Axes
Formatting of Axes• Scaling and aspect ratio of your axes is important to
bring out the relevant relationships and patterns
• Axes should have consistent labeling – no interval jumps
• Breaks in axes should be clearly indicated
Don't Do
Use a Correct Y-Axis Range
Don't Do
Select the Correct Aspect Ratio
D
Don't Do
Properly Formatted Lines
Lines
• Use reference lines (grids) only when an important values should be seen across an entire graph
• Tick marks should generally face outward
• Don’t clutter the interior of the scale-line rectangle with legends, labels, and lines
Grid Lines Can Obscure Data
Don't Do
Proper Use of Color
Color Choices
• Use color only to serve a specific communication goal
• Use more than one color only when the data means different things
• Use a mix of soft colors for the data and brighter colors to accentuate specific features
This is What Bad Color Use Looks Like
Quantitative Use of Color
● Color is mostly arbitrary, that is, non-directional● The human eye cannot perceive quantitative
differences in hue or color
What do the different shades of green mean in reference to the red?
Color Perception Can Be Tricked
All five of these grey boxes are the same shade
Use one color per series
Don't
Do
Color Palette Examples
Standard palette for data
For accentuating features and trends
Use Hues to Order Data
vs
Some Rules About Using Color
● To speed visual search
● To improve object recognition
● To enhance meaning
● To convey structure
● To establish identity
● To … symbolism
● To improve usability
● To communicate mood
● To show associations
● To express metaphors
Components of a Good Graph
●Proper data organization (MOST IMPORTANT)●Formatted without default settings●Designed for ease of readability for others
● Correct use of “preattentive stimulus”● Visual Hierarchy● Readable text● Properly formatted axes● Properly weighted lines● Appropriate color (only when necessary)
Resources for Better Graphs
● Chart Chooser– Extreme Presentation: Chart Suggestions
● http://extremepresentation.typepad.com/files/choosing-a-good-chart-09.pdf
– Juice Analytics: Chart Chooser● http://labs.juiceanalytics.com/chartchooser.html
● Color– Color Scheme Designer
● http://colorschemedesigner.com/