© J.F. Campbell UM St. Louis 2015 1 Data Visualization

© J.F. Campbell UM St. Louis 2015 1

Data Visualization

http://www.nytimes.com/interactive/2009/07/31/business/20080801-metrics-graphic.html?_r=1&




Overview

1. Why use visualization?

2. Types of visualizations.

3. Design guidelines.

4. Infographics.

5. Tableau example.

Visualize This http://www.youtube.com/watch?v=mkEXx7sDXAI#t=69

Much of this is drawn from materials at the Duke University Library Data Visualization site: http://guides.library.duke.edu/datavis/

http://www.youtube.com/watch?v=mkEXx7sDXAI#t=69




Why not Statistics?

• Consider the following four sets of 11 (x,y) coordinates:

• Are they similar?

x y

10 8.04

8 6.95

13 7.58

9 8.81

11 8.33

14 9.96

6 7.24

4 4.26

12 10.84

7 4.82

5 5.68

x y

10 9.14

8 8.14

13 8.74

9 8.77

11 9.26

14 8.1

6 6.13

4 3.1

12 9.13

7 7.26

5 4.74

x y

10 7.46

8 6.77

13 12.74

9 7.11

11 7.81

14 8.84

6 6.08

4 5.39

12 8.15

7 6.42

5 5.73

x y

8 6.58

8 5.76

8 7.71

8 8.84

8 8.47

8 7.04

8 5.25

19 12.5

8 5.56

8 7.91

8 6.89

1 2 3 4


Summary Statistics

• Statistically, they seem pretty similar…

Mean: 9.00 9.00 9.00 9.00

Variance: 11.00 11.00 11.00 11.00

Correlation: 0.816 0.816 0.816 0.816

Linear regression line = 3.00 + 0.500X for all 4!

x y

10 8.04

8 6.95

13 7.58

9 8.81

11 8.33

14 9.96

6 7.24

4 4.26

12 10.84

7 4.82

5 5.68

x y

10 9.14

8 8.14

13 8.74

9 8.77

11 9.26

14 8.1

6 6.13

4 3.1

12 9.13

7 7.26

5 4.74

x y

10 7.46

8 6.77

13 12.74

9 7.11

11 7.81

14 8.84

6 6.08

4 5.39

12 8.15

7 6.42

5 5.73

x y

8 6.58

8 5.76

8 7.71

8 8.84

8 8.47

8 7.04

8 5.25

19 12.5

8 5.56

8 7.91

8 6.89

1 2 3 4

© J.F. Campbell UM St. Louis 2015 56 8 10 12 14 16 18 200

2

4

6

8

10

12

14

0 5 10 150

2

4

6

8

10

12

14

2 4 6 8 10 12 14 160

2

4

6

8

10

12

14

Similar?

x y

10 7.46

8 6.77

13 12.74

9 7.11

11 7.81

14 8.84

6 6.08

4 5.39

12 8.15

7 6.42

5 5.73

x y

8 6.58

8 5.76

8 7.71

8 8.84

8 8.47

8 7.04

8 5.25

19 12.5

8 5.56

8 7.91

8 6.89

1

3 4

2 4 6 8 10 12 14 160

2

4

6

8

10

12

142


Why Visualization?

• To discover new things about your data.

– The most effective way for humans to understand complex data (and large amounts of data) is visually!

• To tell a story using data.

• To provoke and answer questions.

• To facilitate analysis.

• To better communicate.

• Visualization leverages human visual capabilities for data analysis.


The Great One

http://dataremixed.com/2011/08/tribute-to-the-great-one/

http://dataremixed.com/2011/08/tribute-to-the-great-o

http://dataremixed.com/2011/08/tribute-to-the-great-o


Stages

1. Identify the topic of interest and relevant questions.

2. Obtain useful and relevant data.

3. Explore the data to identify interesting relationships: Look for trends, patterns and differences across

categories, space and time.

4. Represent the data (maps, charts, etc.).

5. Refine the presentation with your audience in mind.

6. Provide tools to manipulate or interact with the data.


Types of Visualizations

1. 2D and Planar (geospatial):

a. Types: Choropleth, Cartogram…

b. Use a map to show where something is.

c. Maps are best combined with other charts to provide details on what the map shows.

2. Temporal: For changes over time.

a. Time series or line chart.

b. Stream graph.

c. Polar chart.

http://lgimages.s3.amazonaws.com/data/imagemanager/62927/flickr_jason-morrison_2998016737.png


Temporal Charts

http://www.nytimes.com/interactive/2008/02/23/movies/20080223_REVENUE_GRAPHIC.html#

http://www.nytimes.com/interactive/2008/02/23/movies/20080223_REVENUE_GRAPHIC.html

http://www.nytimes.com/interactive/2008/02/23/movies/20080223_REVENUE_GRAPHIC.html


1. Sankey diagram:

Map flows.

2. Histogram or bar chart.



3. Bubble chart.

With motion: http://www.logeeka.com/motion_chart.html


http://www.logeeka.com/motion_chart.html

http://www.logeeka.com/motion_chart.html



4. Tree maps and hierarchical charts.

http://homes.cs.washington.edu/~jheer/files/zoo/ex/hierarchies/icicle.html



5. Networks.

Vaccine game: http://vax.herokuapp.com/game

6. Radar chart. http://worldshap.in/

http://vax.herokuapp.com/game

http://vax.herokuapp.com/game

http://worldshap.in/

http://worldshap.in/


Baseball Visualizations

Spray charts for Justin Heyward

http://www.fangraphs.com/

1.10% 1.40% 2.40% 2.90% 2.90% 2.30% 1.50% 1.30% 1.00% 0.90%

0.20% 0.40% 0.70%

0.30% 0.50% 0.70% 0.70% 1.00% 1.30% 0.70% 0.30%

0.50% 0.90% 1.40% 1.90% 1.80% 1.20% 0.80% 0.60%

0.90% 1.60% 2.50% 2.80% 2.10% 1.20% 1.10% 1.00%

2.00% 2.80% 2.90% 2.90% 2.50% 1.50% 0.90% 0.50%

2.10% 2.50% 3.00% 3.10% 2.10% 0.90% 0.50% 0.40%

1.70% 2.30% 2.50% 2.10% 1.20% 0.60% 0.50% 0.40%

1.60% 1.80% 1.70% 1.80% 1.10%

0.5

0.60% 0.40% 0.20%

1.50% 1.10%

Wainwright’s 1st pitch to RH batter: strike=67.4%

batt

er

3.60% 0.80% 1.20% 1.60% 1.40% 1.50% 1.20% 1.20% 1.40% 1.40%

1.00% 0.80% 0.80%

4.30% 6.10%

1.20% 0.80% 0.80%

1.50% 1.20% 1.20% 1.60% 1.50%

1.50% 1.10% 0.80%

1.30% 1.40% 1.10% 1.20% 1.50%

1.50% 1.40% 0.80%

1.50% 1.30% 1.20% 1.30% 1.30%

1.20% 1.00% 1.40%

1.20% 1.50% 1.50% 1.40% 1.30%

1.20% 0.80% 1.10%

0.80% 1.10% 1.50% 1.50% 1.50%

0.80% 0.80%

0.80% 0.80% 1.20% 1.40% 1.20%

0.80%

0.80% 0.80% 1.20% 1.60% 1.40%

8% 0.80%

Pitch to RH batter with 0-2 count: strike=46.0%

batt

er


Visualizing Wind

http://www.fangraphs.com/

Live: http://hint.fm/wind/

http://hint.fm/wind/




Design

• Design is not just what it looks like and feels like. Design is how it works.– Steve Jobs, 2003

Clarity

Aest

heti

cs

Confusing

Ugly

Beautiful

Clear

Yes

No ?

?


Design From http://vizwiz.blogspot.com/2012/04/nielsens-advertising-audiences-report.html


Visualization Design Guidelines

• The visualization must have a purpose!– All elements should work together to achieve the

purpose.

– What questions can or does it answer?

– What questions should it answer?

• Be simple and succinct.– Show the main points – do not make the audience try

to figure it out.

– Do not present too much information! (Limit a dashboard to 2-4 elements/views).

• Any interactivity should be obvious to the viewer.


Visualization Design Guidelines

• Many visualizations combine several elements (views, charts, etc.) in a “dashboard”.

• Place the most important view at the top, or top left.

• Be sure the legends are associated with the correct view.

– Position legends to the right of the view, if possible.

• If elements are linked interactively, arrange them top to bottom and left to right, with the linking and filtering starting at the top.


Choosing a Good Chart

http://extremepresentation.typepad.com/blog/2006/09/choosing_a_good.html


Design Guidelines: Charts

• Put the most important data on the rows and columns (x and y axes);

• Use color and size for less important attributes.

• Bar charts are usually better than pie charts:

– Areas in pie charts are difficult to estimate, and the eye can compare only adjacent slices.

– Put labels on the bars.

• Do not use 3D charts.

• Make sure all axes are understandable.

– Axis scales must be consistent.

• With line charts, limit the number if lines and highlight the most important line(s).


Line Charts #1

• Keep it simple!

• Label the lines, instead of using a legend.


• Highlight what is important.

Line Charts #2

• Is the baseline 0?


• Elevate the axis if baseline is not 0

Line Charts #3

• Use a good aspect ratio.


Bar Charts

• Use horizontal bar charts, rather than vertical bar charts.


Tables? From http://www.edwardtufte.com/bboard/q-and-a-fetch-msg?msg_id=0000Jr


Color

• Color is important! Choose colors intelligently.– Use at most 6 colors.

– Use no more than two color palettes.

– Use meaningful colors (pink/blue; red/green, etc.), but be aware that colors are culturally dependent

• Avoid multiple schemes.

Some colors do not work well together!!


Color

• Vary the saturation level (lightness), not the hue (color).

• Consider that your visualization may be printed in black and white.

http://mkweb.bcgsc.ca/biovis2012/krzywinski-visualizing-biological-data.pdf


Color Can Be Deceiving…

Which square is darker – A or B?

Which is darker – A, B or C?


More Colors

Which dog is bluer?

How many colors are in this?


100 Points

What do you see here?

Most points are blue, one is red and four are green.

The points are spread out “evenly” over the space.


Differences are more difficult to distinguish with symbols alone.


100 Points Again…

You may not appreciate that one point is very unusual point - both an uncommon color and an uncommon shape (the green square)

Combining color and shape does not work well!


Most points are blue, one is red and some are green.

Some are squares, but most are dots; one is a +.

The points are spread out “evenly” over the space.


Fonts

• Use only a few fonts:

– Verdana or Trebuchet for numbers.

– Arial, Georgia, Tahoma, Times New Roman, Lucida Sans.

• Use a few appropriate font sizes.

• Change adjacent fonts by only one attribute (bold or underline, not both):

– A good change A Bad change


Infographics

• A common type of visualization specific to a particular context.

• Usually created for a single dataset for a particular purpose.

• Not designed for the user to explore the data.

• Most view infographics as a type of visualization; but some see it the opposite way.


Infographic 1


Infographic 2


Infographic 3


Summary

• Use the real estate wisely.

• Show the main points – do not make the audience try to figure it out.

• Do not present too much information!

• Do the squint test:

– What stands out? What do you see?

• Show it to someone else and ask what they see.

• Include the source of the data.


• A great site for visualization basics.

• A great site for Tableau information.

• More design guidance…

Basic Information

http://guides.library.duke.edu/tableau

http://www.youtube.com/watch?v=pD_OvRtH0aY

http://guides.library.duke.edu/datavis/









Baby Names in Tableau

• Consider the top baby name in each US state for each year…

http://www.tableau.com/public/BabyNamesTraining

• What to call on 4th down?

http://datographer.blogspot.com/2014/03/fourth-down.html






Data for Baby Names in Tableau

• Original Data:

• Every baby name used >5 times, by state and by year since 1910.

• State, Gender, Year, Name, # of occurrences

• From this, extract the top male and female name for each state for each year.

AK,F,1910,Mary,14

AK,F,1910,Annie,12

AK,F,1910,Anna,10

AK,F,1910,Margaret,8

AK,F,1910,Helen,7

AK,F,1910,Elsie,6

AK,F,1910,Lucy,6

AK,F,1910,Dorothy,5

AK,F,1911,Mary,12

AK,F,1911,Margaret,7

AK,F,1911,Ruth,7

AK,F,1911,Annie,6

AK,F,1911,Elizabeth,6

AK,F,1911,Helen,6

AK,F,1912,Mary,9

AK,F,1912,Elsie,8

AK,F,1912,Agnes,7

AK,F,1912,Anna,7

AK,F,1912,Helen,7

AK,F,1912,Louise,7

AK,F,1912,Jean,6

AK,F,1912,Ruth,6

AK,F,1912,Alice,5

AK,F,1912,Esther,5


Tableau Dashboard

Number of different top male (blue) and female (pink) names in the 50 states since 1910

Top name in each state for chosen year

Frequency of name (for top names)

Trend of name as the top name in states over time

YEAR

Gender



© J.F. Campbell UM St. Louis 2015 1 Data Visualization

Documents

campbell um

data visualization http

data maps

data analysis

relevant data

complex data

edudatavis slide

large amounts of data