Introduction to Visualization
•Download as PPTX, PDF•
1 like•729 views
A presentation on the keys to good data visualization, trends in the market, and how companies are using data visualization solutions.
More Related Content
Similar to Introduction to Visualization
Similar to Introduction to Visualization (20)
Recently uploaded
Recently uploaded (20)
Introduction to Visualization
- 1. Juice, Inc.©2016 Chris Gemignani Principal chris@juiceanalytics.com Zach Gemignani Principal zach@juiceanalytics.com Introduction to Visualization
- 2. we craft applications that make using data enjoyable and rewarding Juice, Inc.©2016
- 3. Juice, Inc.©2016 Don’t be accidental Have a message Keep it simple Provide context Apply design fundamentals
- 4. Have a message Juice, Inc.©2016
- 6. “Perfection is achieved, not when there’s nothing more to add, but when there is nothing left to take away.” Antoine de Saint-Exupéry Squelch the noise Juice, Inc.©2016
- 8. Juice, Inc.©2016 Show only what’s needed - nothing more
- 9. Juice, Inc.©2016 Leaders Laggers Norms Historical Tendencies Internal drivers External drivers Who’s the model? Who needs coaching? Who’s out of line? Is change happening? What’s driving? Whose related? Align your charts with your intent
- 10. Juice, Inc.©2016 One metric across many categories Relationships across categories Outliers Find leaders, laggers, and anchors by using: Comparisons and Distributions
- 11. Juice, Inc.©2016 Metric for one category across time Who is changing Transitions Find historical tendencies by using: Trends
- 12. Juice, Inc.©2016 Metric with subcomponents Factors driving change Discover internal drivers by using: Compositions Year
- 13. Juice, Inc.©2016 Clustering Outliers Discover external drivers by using: Relationships Many metrics for many categories
- 15. Juice, Inc.©2016 Keep it simple
- 16. Juice, Inc.©2016 ...with ‘simplistic’ ” Garr Reynolds “Don’t confuse ‘simple’...
- 17. Juice, Inc.©2016 Surveying the results of psychological experiments in the 1950s “My problem is that I have been persecuted by an integer.”
- 18. Juice, Inc.©2016 0 2 4 6 8 10 Binary digits Numbers Letters + Digits Short words Long words Number of items we can recall
- 20. Juice, Inc.©2016 2,147,483,648bytes 7digits 5 one syllable words 2 seconds audio 3 visual images
- 21. Juice, Inc.©2016 0 1,000,000,000 2,000,000,000 3,000,000,000 Computer You Working memory capacity
- 22. Juice, Inc.©2016 “Easy is hard” Pat Hanrahan
- 23. Juice, Inc.©2016 Without context: 100 people, 100 conclusions Context gives data meaning
- 24. Juice, Inc.©2016 career home runscareer home runs
- 25. Juice, Inc.©2016 career home runs
- 26. Juice, Inc.©2016 284 +4.5% v. last month -2.8% v. goal comparison to previous time period and goal 0 100 200 300 400 Q1 2009 Q2 2009 Q3 2009 Q4 2009 Goal Actual shown within broader context of overall trend and goals relevant scope in Q3 2009 new leadsmetric name the new “Vortex” marketing campaign drove new leads after a slow Q2 explanation of mitigating factors Types of data context
- 27. Juice, Inc.©2016 People context Chief Marketing Officer Marketing Analyst How does the data fit into my workflow? What expectations or assumptions do I have? How do I interpret the metric? What other information am I looking at? Does it support or contradict? Who is sharing the data? Do they have credibility?
- 28. Juice, Inc.©2016 Treemaps Sparklines Small multiples 3 visualization techniques worth knowing
- 29. Juice, Inc.©2016 See the full picture Biggest segments Movers See detail Treemaps: The cure for big data display
- 30. Juice, Inc.©2016 Spending by category/line item
- 31. Juice, Inc.©2016 Sales by title, retailer
- 32. Juice, Inc.©2016 Sparklines “Small high-resolution graphics usually embedded in a full context of words, numbers, images. Sparklines as datawords: data-intense, design-simple, word-sized graphics.” - Edward Tufte 20.59% Direct Traffic word-sized graphics
- 33. Juice, Inc.©2016 Sparklines add context Comparison of values
- 34. Juice, Inc.©2016 Sparklines add context Trend
- 35. Juice, Inc.©2016 Sparklines add context Highest and lowest values
- 36. Juice, Inc.©2016 Sparklines add context Comparison to a goal
- 37. Juice, Inc.©2016 Step 1 Step 2 Small multiples Create a simple chart Replicate the chart for related parameters using the same structure and scale
- 38. Juice, Inc.©2016 http://www.nytimes.com/interactive/2009/04/16/us/20090419-Agency-Suburbs.html displayed by CITY NY Times storytelling with small multiples
- 39. Juice, Inc.©2016 http://www.nytimes.com/interactive/2008/05/30/sports/playmagazine/20080601_WOODS_GRAPHIC.html displayed by METRIC NY Times storytelling with small multiples
- 42. Juice, Inc.©2016 Right and wrong in data visualization
- 43. Juice, Inc.©2016 6 principles describe how we determine visual context
- 44. 1) Principle of Proximity Things that are spatially close belong to a group Juice, Inc.©2016
- 45. 1) Principle of Proximity Things that are spatially close belong to a group Juice, Inc.©2016
- 46. 2) Principle of Similarity Things that are spatially close belong to a group Juice, Inc.©2016
- 47. 3) Principle of Enclosure Objects that are enclosed by a shape will be seen as a group Juice, Inc.©2016
- 48. 3) Principle of Enclosure Objects that are enclosed by a shape will be seen as a group Juice, Inc.©2016
- 49. 3) Principle of Enclosure Objects that are enclosed by a shape will be seen as a group Juice, Inc.©2016
- 50. 4) Principle of Closure We strive to perceive shapes as complete Juice, Inc.©2016
- 51. 4) Principle of Closure We strive to perceive shapes as complete Juice, Inc.©2016
- 52. 5) Principle of Continuity Objects that are aligned are perceived as a group Juice, Inc.©2016
- 53. 6) Principle of Connection Objects that are connected are part of the same group Juice, Inc.©2016
- 54. Things that are spatially close belong to a group Proximity Similarity Things that share color, size or shape belong to a group Enclosure Objects that are enclosed by a shape will be seen as a group Closure We strive to perceive shapes as complete Continuity Objects that are aligned are perceived as a group Connection Objects that are connected are part of the same group Juice, Inc.©2016
- 55. Increase data-ink ratio (make every pixel tell a story about your data) Fundamental rule of chart design Reduce chartjunk (remove chart elements that are decorative or ornamental) Juice, Inc.©2016
- 56. Remove chart-junk Images courtesy of Tim Bray: http://www.tbray.org/ongoing/When/200x/2003/03/13/Data-Ink Increase data-ink Increase data-ink Juice, Inc.©2016
- 57. Fundamental rules of table design Increase data-ink ratio (make every character meaningful) Reduce tablejunk (remove chart elements that are decorative or ornamental) Juice, Inc.©2016
- 61. Objects that are aligned are perceived as a group Principle of Proximity Things that are spatially close belong to a group Principle of Continuity Juice, Inc.©2016
- 64. Increase data-ink ratio vJuice, Inc.©2016 Juice, Inc.©2016
- 65. Increase data-ink ratio 1 degree latitude = ~60 miles .001 degree latitude = 300 feet 0.00000000000001 degree latitude = 1 micron, the width of a very small bacteria Hence, the software must have been pirated by a bacteria Juice, Inc.©2016
- 66. Increase data-ink ratio Juice, Inc.©2016
- 67. Increase data-ink ratio Juice, Inc.©2016
- 68. Increase data-ink ratio Juice, Inc.©2016
- 69. Tables Remove gridlines Use lines or whitespace to separate areas that are conceptually different Display the smallest amount of numbers that you can to support the needs of the table Juice, Inc.©2016
- 70. Contrast Use contrast to draw attention Juice, Inc.©2016
- 71. Copyright 2010 Stanford Visualization Group Contrast Juice, Inc.©2016
- 72. Less effective You can use color to emphasize text. You can bold-italics your important text. You can UPPERCASE your important text. You can underline your important text. Emphasis You can add italic letters. You can bold your important point. You can use color to emphasize text. You can combine effects. Juice, Inc.©2016
- 73. Capital letters MAY I CRAVE YOUR INDULGENCE TO OPEN THIS BUSINESS DISCUSSION BY A FORMAL LETTER OF THIS SORT. IT IS PERTINENT THAT BUSINESS OF THIS MAGNITUDE SHOULD HAVE COMMENCED PROPERLY WITH A FORMAL MEETING OF YOU AND US TO ENABLE US KNOW OURSELVES, HAVE A FORE KNOWLEDGE OF THE NATURE OF THE BUSINESS, DISCUSS AND ACQUAINT OURSELVES WITH THE RESPONSIBILITIES AND FUNCTIONS OF PARTIES CONCERNED, AND APPROPRIATE SHARES ACCORDINGLY. HOWEVER, FOR TIME FACTOR, CONFIDENTIALITY AND PERSONALITY OF PEOPLE INVOLVED HERE IN NIGERIA, WE CHOSE THIS APPROACH FOR THE REMITTANCE OF FIFTEEN MILLION, FIVE HUNDRED THOUSAND U.S DOLLARS (US$15.5M), PLEASE BEAR WITH US FOR MAKING THE INITIAL CONTACT THROUGH E-MAIL. BUT MY PARTNERS ARE INSISTING FOR A MEETING IN ORDER NOT TO FALL INTO A WRONG HAND AGAIN. Juice, Inc.©2016
- 74. How we read Aoccdrnig to rscheearch at Cmabrigde uinervtisy, it deosn't mttaer waht oredr the ltteers in a wrod are, the olny iprmoetnt tihng is taht the frist and lsat ltteres are at the rghit pclae. The rset can be a tatol mses and you can sitll raed it wouthit a porbelm. Tihs is bcuseae we do not raed ervey lteter by it slef but the wrod as a wlohe http://blogs.msdn.com/fontblog/default.aspx?p=2 Juice, Inc.©2016
- 75. THE QUICK BROWN FOX JUMPS OVER... The quick brown fox jumps over ... Juice, Inc.©2016
- 77. Juice, Inc.©2016 Reds Whites Hearty, earthy, jammy, rustic Tart, sophisticated, citrusy, light Juice, Inc.©2016
- 78. Juice, Inc.©2016 Serif Sans-serif Hearty, rustic, somber, dignified Light, modern, smooth, simple Serif Serif Juice, Inc.©2016
- 79. Juice, Inc.©2016 Reds Whites Merlot, Shiraz, Cabernet Sauvignon, Pinot Noir Chardonnay, Sauvignon Blanc, Riesling Juice, Inc.©2016
- 80. Juice, Inc.©2016 Serif Sans-serif Georgia Arial / Helvetica Serif Serif Verdana / Tahoma Times New Roman Palatino Trebuchet Juice, Inc.©2016
- 81. Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum. History of Medicine Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, Fonts - Simple rules Juice, Inc.©2016
- 82. Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum. History of Medicine Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, Body Text Clean readable text, 50-80% of your text will look like this. •10-16pt •Arial or Georgia Juice, Inc.©2016
- 83. Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum. History of Medicine Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, Header Use to separate and name sections •150%-200% of body Juice, Inc.©2016
- 84. Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum. History of Medicine Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, Notes Additional things a user should be aware of •85% of body •Lower contrast Juice, Inc.©2016
- 85. Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum. History of Medicine Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, Emphasis Draw the eye to key points •Same size as body •High-impact color/bold/italic Juice, Inc.©2016
- 86. Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum. History of Medicine Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, Emphasis Draw the eye to key points •Same size as body •High-impact color/bold/italic Juice, Inc.©2016
- 87. Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum. History of Medicine Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, Juice, Inc.©2016
- 88. Categorical Lots of contrast between each adjacent color Sequential Colors can be ordered from low to high Low High Diverging Two sequential schemes extended out from a critical midpoint value Low HighCritical midpoint http://colorbrewer2.org/ Color schemes for data Juice, Inc.©2016
- 89. Sequential Colors can be ordered from low to high Low High Juice, Inc.©2016
- 90. Diverging Two sequential schemes extended out from a critical midpoint value Low HighCritical midpoint Juice, Inc.©2016
- 91. Categorical Lots of contrast between each adjacent color Juice, Inc.©2016
- 92. Juice, Inc.©2016 Have a message Keep it simple Provide context Apply design fundamentals
- 93. www.juiceanalytics.com @ juiceanalytics Chris Gemignani twitter: chrisgemignani chris@juiceanalytics.com Zach Gemignani twitter: zachgemignani zach@juiceanalytics.com Juice, Inc.©2016
Editor's Notes
- Welcome to the Juice Analytics [you are here] Event! * Thank you. I’m Ken Hilburn and this is Ron LeSieur and we work with Juice Analytics. Thanks for coming. We’re doing this mini tour to help make it easier for everyday folks to communicate better with data. We’ve gotten great feedback on this series and we really think you’re going to find what you see today enjoyable and useful. At the end we’ll be raffling off some goodies, so hopefully you can hang around until the end. Also, we’ll be sending out the slides to all who registered.
- [for Tableau] Let me see a show of hands: How many folks here were familiar with Juice before you walked in the door? Juice works with clients of all sizes to help them take the data that they provide to their customers and make that information speak loudly and clearly, driving value into their ecosystem. For those of you who have been following Juice for a while, you will know that we put a lot of effort into investing back into the information worker community to make using data enjoyable and rewarding for everyone, not just analysts. We believe there is vast untapped value in data that can be released with proper information design and visualization. When properly done, visual representation of data can be truly powerful. To demonstrate that, I’d like for you to watch a short video.
- Today is about communicating with data in a clear and intuitive way. And we’re going to cover 4 areas that will help you be more intentional. Individually, the things we’ll show you aren’t terribly complex. But together, we think they can make a big difference helping you in your work. Four simple things we’d like you to walk out of here with: * First, be clear about what you’re trying to communicate. Dont’ leave your audience confused about your point * Second, when you are presenting data, don’t try to get too sophisticated. Keeping things simple can make a world of difference in whether your audience “gets” it. We are big proponents of “simpler is better.” * Third, account for the context of the data and your audience to connect with what’s important to them * Finally, proper application of design principles can reduce barriers to understanding.
- First: start with your message. Having a message is what makes it worth while to listen - just like with this presentation; you guys are hoping we have a message and not just a bunch of points. When we talk about data, the data isn’t the message. The message is the truth revealed by the data. When you present data, it’s your responsibility to make the message speak on the data’s behalf. Be crystal clear about your message and don’t trust the data to speak for itself.
- Who knows what the message is for this dashboard? Of course not. There is no message. There’s lots of great information here, shown in some pretty cool ways. But what is it helping me do? Or is it actually leaving me more distracted and confused than before I looked at it? This is the same problem that many information presentations suffer from.
- It’s quite interesting to present as much information as we can. However, let’s not confuse interesting with actionable. Hone your message and focus on the things your audience can act on, not just those they can think about. People love data visualization when it helps them get stuff done. You know, here’s a trick I use to help squelch the noise. When someone shows me some data related item, I frequently ask a very simple question “so what? What is your expectation regarding the action I should take as a result of knowing that information?” Keep in mind that “Interesting” doesn’t cut it. Keep asking yourself “If you knew, what would you do?”
- If you browse to goingtorain.com, it will tell you in a very simple manner “yes” or “no” based on your location. They’ve remained ruthlessly committed to their message. When look at data in our businesses, isn’t this really what we’re trying to get to? Should we hire another developer? Are we going to meet our sales objectives? Will the delivery be on time? Start with your message in mind and make sure, if you don’t do anything else, that you deliver that message. So, let’s look at a more practical example [click]
- Here is a dashboard we created for a client that identifies and reports on internet network congestion. As you can understand, there are 10s of thousands of measure points in this sort of system, but the approach we took was to simplify down to simply how much capacity is being used across the entire network which is represented by the stacked thermometer charts. For this dashboard, there are only three metrics they care about: capacity, utilization, and congestion. Everything else is supporting information.
- When we have a message, we start to form questions that we that need to be answered to support it. And for each of these factors, like drivers, or norms, or trends, there are chart types that are specifically suited to revealing important information.
- So, let’s talk about a few specific applications of proper chart styles using some data about the population in counties in Ireland between 1841 and 2002. In this special stacked bar chart (called a bullet chart) we see clearly that Cork had the largest population in 1841 and Dublin in 2002. This class of chart is called a comparison or distribution chart. [Click] This type of chart highlights relationships across categories and outliers.
- Here we see similar data shown for each county over time. Again we see information about Cork (the red line) and Dublin (the purple line) that better represents what has happened over time. These sorts of charts are clearly called “trends”. [Click] This type of chart hightlights transitions and who is changing.
- If you look at the total population of Ireland, you might be interested in how that number is comprised by each county. A stacked or “composition” chart shows this well. [Click] Composition charts, show components that make up a metric, but when used with trend lines, can show factors that drive change.
- And finally, by plotting two different metrics together in a scatter chart, you can see how different metrics relate to and influence each other. [Click] These charts can give you a feeling for the outliers as well as clustering. This class of chart is called a “Relationship” chart.
- So, you may say “that was cool, but how do I remember all those chart types again?” Well, the good news is, you don’t have to. We’ve created a free tool that helps you select the right chart based on what you’re trying to do. Just go to chartchooser.com and select your filter. But wait, there’s more: this tool allows you to download excel and power point versions of the charts for your task.
- As Ken said, just because you have the data, doesn’t mean you need to show all of it. Nevertheless, it is very easy to go overboard by showing too much data -- and in ways that are hard for people to understand In this section: * Examples of things to avoid and * Visualization techniques that can take things to another level while still keeping things simple. Key point: if you find yourself trying to explain how a chart works, you’ve lost.
- So, let’s talk a little a little more about how to bring the right kind of simple into your information presentation. I want to start by talking about a paper that’s been influential in how we think about human cognition. George Miller surveyed the results of psychological experiments in the 1950s. At that time, people were looking at what we can store in our short term memory. The results may surprise you.
- Our short term memory is extremely limited. Without specialized training, we can repeat back about 9 binary digits, 7 numbers. These values go down as concepts get more complicated.
- The computer you’re using right now probably has a memory capacity of around 2 gigabytes, that’s a little over 2 billion bits
- Your computer's short term memory capacity 2,147,483,648 bits (+-1073741824) Your brain 7 (+-2) 2 seconds Or we can show that in a graph....
- So how do we accomplish anything with these little tiny pea brains that we have? One way we do this is our brain is constantly attuned to patterns. Our brain hungers to find patterns in our world.
- Context is understanding the bigger picture; without context, everyone will simply apply their own understanding to arrive at conclusion The problem with relying on intuition to interpret information is that some people are better at it than others. This leads to different conclusions based on the factors that influence them and their understanding of the data. Proper understanding of context removes the variability that comes from intuition. Both people and data have separate but important context that enable better interpretation
- what does a number mean here’s a complicated one to illustrate my story
- it can be complicated
- Data is only meaningful when presented with the contextual relationships it has. The challenge is presenting all this information/context, without
- Not just data; data that is meaningful based on your audience’s needs and abilities People look at the world from the perspective of their job role, their relationships, their location, and their experience. To fully connect with them, you need to understand their context and present not just data, but data that translates into meaningful information in a way that fits their needs and ability
- There are three visualizations that we feel qualify along all three of these criteria. [By a show of hands, how many people are familiar with treemaps? Sparklines? Small multiples?]
- Raise your hand if you are familiar with this visualization? [audience participation] It is call the Map of the Market and it shows market sectors and individual companies * size of each box represents market cap * color of each box represents the change for the day. Treemaps aren’t exactly simple They are one of the most effective ways to show complex data Here’s what they do well: 1. Show the full picture 2. Zoom down to detail 3. See what is big...and therefore important 4. See what is changing, up and down.
- And it is surprisingly versatile. We’ve found a lot of situations where a treemap adds value. One question we get all the time is: Do people “get” treemaps? Isn’t there a risk of alienating users?
- Now that’s hardly big enough to be called a visualization, wouldn’t you say? What’s the point? To provide graphical context to a number: For example, a sense of the trend or comparison to other numbers.
- They offer a bit of visual guidance, context to a set of numbers. There are a number of ways you can use sparklines: * Simple way to make it easier to pick out the biggest and smallest values in a long table * Show the trend for a value -- have things been going up, down, a bit of both? * Trend can get more sophisticated by showing highs and lows * Alerts, and how the value fits relative to a goal
- The small multiples approach to this problem is to show that same simple map, but for different cuts of the data First, we can see how the map changes by income level Then we can see the different maps by income level and ethnic group. Suddenly there are some great stories that are exposed. And for the audience, they just need to understand a simple chart, then look for how that chart changes across different dimensions.
- The great thing about building blocks is that if you know how to use them, you can build some pretty amazing creations. So, let’s talk about visualization building blocks.
- [pause] Everybody has preferences about what they think is attractive, right? So why do we think that looks don’t matter when it comes to data visualization? Let’s take a look look at a few ways to make your data look like a fuzzy kitten.
- Data visualization isn’t entirely subjective -- there are things that are right (click) and things that are wrong (click) Good visualizations communicate information effectively; bad visualizations can confuse, waste time, or even be deceptive. Let me cover a few of the common mistakes
- The logic that lets us decode captchas can be summarized in 6 principles. Many of these are self-evident and we’ll go fast. how the brain processes visual context
- The principle of proximity. We perceive things that are close together as belonging to a group [Click] For instance, we see this as three groups, rather than 10 dots.
- This is very useful in tables, because we can accentuate whether or now we want users to focus on horizontal comparisons which we can accomplish by spacing the data out more vertically, Or, we can emphsize reading down the columns by spacing the data out more horizontally.
- This is very useful in tables, because we can accentuate whether or now we want users to focus on horizontal comparisons which we can accomplish by spacing the data out more vertically, Or, we can emphsize reading down the columns by spacing the data out more horizontally.
- There’s nothing inherent in the data to suggest this.
- Objects enclosed by a shape are seen as belonging to a group. We see this as a grid of four graphs, even though if we removed the boxes
- Objects enclosed by a shape are seen as belonging to a group. We see this as a grid of four graphs, even though if we removed the boxes
- We see this shape as an oval, rather than a curve
- Under what conditions do we perceive these two faint lines as a box. We just need a little help. [Click] Some data in there gives our eyes a hint. [Click] And we perceive this as a box.
- Finally, the principle of connection Objects that are connected are part of the same group. We might not see the value in these dots, but once we connect them with lines [Click] we see the trend. None of these principles will surprise you. I hope not, you’ve been using them for years. But what you may not have thought of is applying them to help your audience see data efficiently. Think back to our practical Excel skills video, where we had a graph like this.
- Finally, the principle of connection Objects that are connected are part of the same group. We might not see the value in these dots, but once we connect them with lines [Click] we see the trend. None of these principles will surprise you. I hope not, you’ve been using them for years. But what you may not have thought of is applying them to help your audience see data efficiently. Think back to our practical Excel skills video, where we had a graph like this.
- None of these principles will surprise you. I hope not, you’ve been using them for years. But what you may not have thought of is applying them to help your audience see data efficiently. Think back to our practical Excel skills video, where we had a graph like this.
- The primary rule of chart design is to look at every element in your chart and make sure it is supporting the story the data is telling. These rules were popularized by Edward Tufte (professor emeritus at Yale). Let me show you an example of what is possible when we remove ornamentation
- In this example, first the extraneous decoration is removed, then we made sure that all the pixels were working hard.
- These are basically the fundamental rules of table design too. We remove elements that are decorative and make sure that every character (every letter or number) is meaningful.
- Here’s a table. Where’s the tablejunk here? the unnecessary and distracting decorative elements.
- The gridlines and backgrounds are not necessary. Let’s take them away.
- Why can we get away with removing gridlines? Their purpose is to help us see the rows and columns, right?
- Remember our principles of continuity and proximity And after all
- But we do want to add some detail back [click] to make it even easier to see the structure
- We use lines to separate areas that are conceptually different. We could also create this distinctions with whitespace. So, I put a thin medium grey line separating the headers from the body and the body from the summary. A little bit of white space every five lines make it easier to track horizontally. Now what can we do about the data ink ratio?
- Most of these numbers are not analytically useful to us. There’s too much precision, and that precision actually hides the trend from us. In most cases, three digits of precision are sufficient [Click]
- Here’s an example that I found on the Internet. This company measures software piracy and shows a table where piracy is occuring. Can you spot the problem? [4 seconds] Anyone see the issue with this number? PRecision: one, tow three, four... oh what the heck; 14 digits of precision. So, with a few simple calculations, you can only draw one conclusion: bacteria have taken over the internet.
- Most of these numbers are not analytically useful to us. There’s too much precision, and that precision actually hides the trend from us. In most cases, three digits of precision are sufficient [Click]
- This makes the trends clearer, and we can tighten up the table.
- Display the smallest amount of numbers that you can to support the needs of the table. Also notice that we put a header in the top to show the units. I want to talk a little bit more about data precision
- Let’s talk about how to organize data in a table. Generally you want to support a pattern of vertical scanning, followed by horizontal comparisons.
- That means you’ll have a category and subcategory vertically and dates horizontally or
- The table we’ve been examining for Major Leaguye baseball salaries is like this
- Or a category subcategory and metrics associated with each category horizontally
- Here’s an example of a table that has categories and subcategories and multiple metrics
- Let's recap Tables allow us to perform precise lookups and comparison between small numbers of values. Graphs allow us to see the trend, the shape of data. Let's talk about design tips for tables, how can we help people get more.
- Next, let’s talk a little about contrast. Contrast is used to highlight key points in your information. In practice, this is more often about about de-emphasizing the stuff that isn’t as important. Let’s look at an example. [click]
- There’s a lot of data here. But if don’t want to remove the resolution and context that the extra data adds [click] we can deemphasize that data that’s not directly tied to our message.
- The primary way to use contrast is with emphasis. Here are ways of emphasizing text. These are generally in order of increasing emphasis. [PAUSE] [CLICK] Here are a few methods that are less effective. With color in particular you need to ensure there is adequate contrast.
- Use of “all caps” is well suited for importance business correspondence such as this request to transfer $15M from an account in Nigeria... Just kidding. Avoid extended capitalization. Hey, wait a minute... Maybe this was capitalized on purpose.
- Let me tell you some research about how we read. According to research at Cambridge University.... Wait a second. Actually, it’s amazing how quickly we can read this paragraph, given that very few letters are in the right shape. We recognize words through a combination of letter recognition and word shape recognition. The words in this paragraph all have the right starting and ending letters and they have approximately the correct word shape. This leads to one interesting conclusion...
- With capital letters, we do not have a distinctive word shape. All caps is less legible than regular text and generally should be avoided. On the other hand, allcaps
- Let’s spend a few minutes talking about the effective use of Fonts. Simply put, [CLICK] fonts are [CLICK] like wine.
- In the world of wine, we have reds and whites.
- The world of fonts is similarly bisected into serif and sans-serif fonts. Serifs are like reds. Sans-serif are like whites. Serif is the name for these little decorative feet that come off the end of serif letters. Sans is french for “without” so sans-serif simply means, without serif.
- Red wines have a number of major varieties like Merlots, syrah’s and so on Whites are the same.
- Here are some of the fonts that are broadly available. Among serif fonts, Georgia, Times New Roman, and Palatino are generally available. Think of these as the Merlot, Cabernet Sauvignon and Pinot Noir of fonts. Georgia is a lovely font that works well on the screen, at different sizes and in bold and italics. It’s my editors choice. Palatino can be quite elegant, while Times New Roman is workmanlike. On the sans-serif side, some common fonts are Arial (and a very similar font called Helvetica), Trebuchet, and Verdana and Tahoma. Arial is my editors choice. Trebuchet is a sassy font, take a look at that e, quite usable. Verdana and Tahoma are cousins. Tahoma is a just a horizontally compressed Verdana. Think of Verdana as the fat one and Tahoma as the skinny one. They’re very readable.... [pause] Yes, that is a back-handed complement. [pause] Arial! editors choice.
- So, how do you ever select the right one? Many times when you try to build a text based information document, you just end up with a Frankenstein of sizes and fonts. That’s why Juice created we’ve created something we call the simple font framework. In four easy steps you’ll be able to convert your Frankenstein into a Venus Demilo
- We start by specifying the body text. This is what the vast majority of your text will be written in. You want to pick a small-ish size of a simple font like Arial or Georgia.
- Next, use headers to separate and name sections. For headers, base this font on the body. If your body font is 10points in size, your header font would be about 15 to 20 points.
- Often times there notes that you need to add to summarize content or highlight advanced topics. You want notes to be more subtle than the body, so go with the same font style at about 85% size with a lower contrast color.
- Finally, for those few situations where you want to highlight some text, apply the emphasis rules we discussed earlier.
- And now you know how to apply the Juice Simple Font Framework to tighten up your text so that it looks intentional and not accidental.
- Finally, let’s talk just a little about colors. Proper color application can be tricky, but start with knowing about 3 ways to color your data. The first is called Sequential coloring. This is used for data that has an order like quartiles or ranges. The resulting colors therefore also have an order The second is called Diverging coloring. This model is used when there is a mid point or norm in the data. The two color on the ends show extremes. The third is is called Categorical. Use this model when the data is for different items that don’t imply sequence, such as departments, or states. So what are some great color sets to use? Cynthia Brewer has created a great site called colorbrewer that has many awesome color sets that you can use. That was our last building block for today, so with that, I’ll turn it back over to Zach.
- This is the kind of stuff that we get really passionate. And we love to share our thinking...and hope that we can help people get a little better at communicating data. The next time you are putting together an analysis for presentation or reviewing someone else’s work, you’ll keep these three things in mind.