presented at FITC Toronto 2018
More info at http://fitc.ca/event/to18/
Presented by
Corey Ouellette, Thomson Reuters
Overview
When you think of “data visualization” what is the very first thing that comes to mind? For many, it’s bar graphs, pie charts, and histograms, or maybe some combination thereof. You’re not wrong – but it’s so much more than that. The era of pie and bar charts has come and gone; these traditional visualizations alone are insufficient. Now is the time of data visualized on a rich canvas. A canvas that not only informs, but immerses you in information in much the same way that your favourite book immerses you in its narrative.
Objective
When attendees leave, that they walk away with an understanding of how development, design and data are strongly intertwined with one other. When aligned with customers needs, these aspects create a meaningful and actionable experience.
Target Audience
Designers and developers interested furthering their appetite for visualization
Five Things Audience Members Will Learn
How data visualization can lead to data exploration
Creating an experience with information
New models of data visualization
Telling a story through data
How to blend design and development through data visualization
3. THREE HORIZONS
• Data Experience Developer at Thomson Reuters Labs
• Avid kayaker/fisher
• Can speak mandarin
4. • Exploring your data
• Understanding data types
• Visualization choices and considerations
• How data visualization can lead to data exploration
• Creating an experience with information
• New models of data visualization
• Telling a story through data
TALK OVERVIEW
PROCESS
EXECUTION
7. • Is it statistically accurate?
• What is your sample size?
• What is the margin of error?
• Are you looking at a long enough timeframe?
• When you use a sample to represent an audience, you must make
sure that the people in your sample are representative of the
audience.
• Always spot check your data when combining multiple datasets for
errors
• Examine outliers – nuggets or inaccuracies
HOW IT WORKS – IS IT ACCURATE?
8. •Unstructured text vs. structured text
•File format (csv, tsv, txt, json…)
•Encoding( vs. )
HOW IT WORKS – IS YOUR DATA CLEAN?
9. •Gaps in time
•Are you exploring all areas of where you can get data?
•Can you aggregate other data sources
• This can evolve over time
HOW IT WORKS – ARE THERE MISSING PARTS?
11. • What's your hypothesis?
• Are there trends in the data?
• Start initial visualizations in excel/simple graph tools
• Are there outliers?
• Combine datasets
• Corresponding data that would be of value for comparison
• Deeper analysis
• Sentiment analysis
• NLP
• Consider viewing data with different diagrams/visualization
DATA EXPLORATION
12. •Not all datasets are complete
• Data cleaning will never get you to 100%
•Some information is confidential
•Consider ways to show missing data
EXPLORING YOUR DATA – WHY IS THERE
MISSING DATA?
13. Exploring your Data – Are you twisting the
narrative?
• Hiding the greater picture
• Only focusing on a specific time can leave out
important comparison information
• Excluding data points to give greater merit to a topic
• Skewing Visualization in favour of your narrative
• Visualizing data can tell false truths when information
is not being accurately displayed
EXPLORING YOUR DATA – ARE YOU TWISTING
THE NARRATIVE?
19. • Open-ended response
• Tweets
• Address information
• Form data
• Text patterns
• *Sentiment analysis
• *Text extraction
STRING /RESPONSE TEXT
When to use:
• Word cloud
• Text snippet
• Tooltip
• Results output
21. • Longitude and Latitude
• IP address
• Line interpolation
• Directions
• Traffic data
SPATIAL / GEOGRAPHIC
When to use:
• Choropleth maps
• Time lapse points
• Map interfaces
• Event detection
• Point to point travel
22. • API access to consolidated data
• Twitter
• RSS
• News feeds
• Analytics data
• Survey data
• Databases
• Customer data
GROUPED DATA SETS
When to use:
• Dashboards
• Reports
• Customer analytics
25. Index all geo-located tweets referencing #UX in North America over a
few months and explore the information for insights with the UX space
Scenario
27. VISUALIZING DATA - CONSIDERATIONS
• Add radius weighting to follower count
• Sentiment analysis of feeling associated with tweet
28. • Add tooltip to show content of tweets
• Group sentiment to show distribution
29.
30.
31. Index all geo-located tweets referencing #UX in North America over a
few months and explore the information for insights with the UX space
#UXploration of Data
Recap Scenario
33. • Who are your audience?
• How much detail do they need?
• What is the margin of error in your data?
• Is the insight actionable or just informative?
• What story does the data tell?
• Why do you need a visualisation?
VISUALIZING DATA - CONSIDERATIONS
34. • Test different ways of visualizing your data
• Is this for analysisor for story telling?
• Multiple ways of seeing the same information can help reinforce
• Consider the scales and dimensions on what your visualizing
• Leverage the use of *colours
VISUALIZING DATA – HOW TO?
35. • Clear
• Specific
• Keep it simple
• To the point
• Inline with audience
VISUALIZING DATA – CONSIDERATIONS
36. • 3D – novelty vs. insight
• Visualizing data for the sake of visualizing
(number dressing)
• Over complicating the information
• Avoid graphical distortion – pick the right
scale
• Too many colours (no more than 6)
• Reduce the need for math
VISUALIZING DATA – THINGS TO AVOID
38. • Who are your audience?
• How much detail do they need?
• What is the margin of error in your data?
• Is the insight actionable or just informative?
• What story does the data tell?
• Why do you need a visualisation?
VISUALIZING DATA - CONSIDERATIONS
49. • Who are your audience?
• How much detail do they need?
• What is the margin of error in your data?
• Is the insight actionable or just informative?
• What story does the data tell?
• Why do you need a visualisation?
VISUALIZING DATA - CONSIDERATIONS
50. NEW MODEL OF DATA VIZ
Static to Interactive to static and so on.
61. • Exploring your data
• Understanding data types
• Visualization choices and considerations
• How data visualization can lead to data exploration
• Creating an experience with information
• New models of data visualization
• Telling a story through data
IN CLOSING…
PROCESS
EXECUTION