Today (July 18, 2012) the Guardian Data Blog asked the question, "Which are the laziest countries on earth?". The data they used is from the British medical journal The Lancet. To visualize the data, the Guardian used Tableau Public. You can see the interactive work below:
This visualization uses a chart type dubbed a 'Lollipop Chart' by Andy Cotgreave when he worked with the DataStudio (). Andy identified the features of this techique this way:
1. Can be used when all dimension members have high values (i.e. long/tall bars in a bar chart)
2. Greatly reduces the data-ink ratio while maintaining a clear link to axis labels
3. All the users I’ve shown it to so far have really engaged with it – they think it’s both pretty and easy to read
I also like the fact that it works if you add more dimensions to make small multiples:
The first question I had when I saw the Guardian's outing of the laziest countries was, "How do countries within a region stack up against each other?" I also wondered if gender sorted differently for the countries within a region. Then there was the whole issue of averages and variability. As it was published, the Guardian visualization didn't get me there for any of these questions.
One of the hurdles for me was the lollipops themselves. In this example, they take up way too much space for my liking. So, I downloaded the workbook from Tableau Public and made some changes.
One change I made was to use a dot plot instead of a lollipop chart. This saved two lines for each country pane, and showed me more quickly the differences by gender within a country. I also added a filter for the region, so I could see all the countries in a region at the same time instead of having to scroll and remember. Thirdly, I added a parameter to make sorting flexible and interactive. Finally, I added a multi-dimensional strip plot so I could see the distribution of values for each country, one gender at a time.
Combining all these changes together results in the following interactive exploratory analytical tool. Play around with it and let me know if you get more insight about the laziest countries.
Earlier today I received a message to look at this "cool pie chart" from the HIN.com web site. The chart is supposed to show the percent of respondents to a survey who answered a question about actions ERs could do to reduce repeat ER visits by recently discharged patients. They gave away this image for free, encouraging readers to copy and use it in their own blogs or web sites, with complete attribution. Or, you could buy the report that includes this chart for $127.
HIN's Reducing Avoidable ER Visits e-survey conducted in October 2011 captured how 134 healthcare organizations are working to staunch the flow and expense of avoidable ED use and point low-acuity patients in the direction of appropriate care. According to survey respondents, some effective strategies to reduce avoidable readmissions by the recently discharged are:
· Phone follow-up within 2 days: 49 percent
· Primary care physician visit within 3-5 days: 44.9 percent
· Medication review: 38.8 percent
· Home visit: 34.7
· Notify primary care physician of discharge: 30.6 percent
Not good. Don't do this.
Pies are for dessert (apple or pumpkin, please), even if there are only two slices (one for me - say 80%, and one for you - say 20%). Otherwise we are going to fight over whose piece is bigger, because the human brain doesn't visually process angles well at all.
In this case though, the larger offense to the viewers and their brains is that you are supposed to suspend what you know about pies (all the slices taken together are supposed to equal the whole pie - 100%). Here the pie is apparently larger than the sum of its parts. What the author is really trying to convey, we think, is that the question was multi-select, so respondents could choose all that apply. So it is not a parts to the whole problem. To convey the percent of respondents who answered each choice, an ordered bar chart does the trick, or simply a sorted text table. Here is how an ordered bar chart might look. It conveys all the information or the original with no cognitive load (extra processing effort) in a smaller space. And the little Best Practice threshold parameter let's the author and consumers of the data clarify their POVs interactively if necessary. It is clear which interventions are more recommended than others, and by how much -- for example, a phone follow-up at discharge had nearly twice as many mentions as offering a provider appointment at discharge. We can deduce this even without the value labels, because the human brain is much better at deciphering differences in length.
Sometime this month, I have to travel to NYC, and will need access to WiFi. Hopefully, it will be free. But where to start looking? Well, the NYC Open Data program happens to have a 2010 vintage list of WiFi locations in the whole city. Each entry has complete name and address information, along with lat/long. Free vs Fee is another one of the dimensions of the data. Here is what I did with the data in short order, using Tableau Public.
The data set includes a City field, which I thought would have five entries, one for each Borough. However, it seems to be a collection of the Borough names along with neighborhood names. The city 'New York' is really Manhattan Borough, which is of interest to me for this trip. There are quite a few listings there. The other Borough names are in the list of Cities, but contain few locations each, serving, I suspect, as a catch all when no neighborhood is appropriate.
Saturday, June 2, 2012, ESPN More Sports' Stats & Info site carried a special look at Johan Santana's history making no-hitter in a New York Mets - St Louis Cardinal's match-up. The story included a playing field visualization of each out in the game, using Tableau Public. The visualization places a circle at the location of each out. The circles are color-coded by type of play (fly out, ground out, strike out). The inning in which an out occurred is encoded by varying the size of the circle; all outs in inning one supposedly one size, inning two, slightly larger, and so on). Take a look at the image below:
Nice work, but I felt a need for more interaction. What if you want to see quickly all the outs in a particular inning - not easy or simple enough. The hold-over tool tip requires too much effort to decode. The color encoding in the original conforms to Tableau's automatic best practice, but a more color-blind friendly palette might be as effective. Finally, the size encoding is based on the use of a numeric measure for inning, 0.0 to 9.0, when really these are discrete values (more of a dimension/category).
So, I did a rework, shown below. It includes a simple box score, so to speak, identifying each out in each inning, by type. Now the viewer can immediately see how the inning went for Santana, with almost no cognitive load. The box score has instructions telling the viewer to select an inning header to highlight all the inning's outs on the playing field viz. Better interaction. Also, you can select more than one inning, and compare them to one another. The size encoding on the field view is improved for the innings. Size is not really the best encoding here, because the larger circles might suggest "bigger" or "more" or "better", which is not the case here, but it works in a pinch.
When I showed this rework to the author, his comment was, "all we need now is an asterisk for the disputed foul-ball call at 3rd base in the 6th inning." So, back to the authoring board I went to take care of that. For the disputed call, a conditional text field makes a footnote appear in the tooltip when you hover over the box score or playing field marks for the groundout; it reminds that the previous pitch resulted in a hit that was called foul, but replay showed was fair. The footnote does not appear when you hover anywhere else. At the same time, I improved the tooltip, so that all the information one needed was in a simple color coded sentence, based on the play type. This kind of tooltip flexibility is very easy to do in Tableau and is widely applicable.
And baseball fanatic colleague Matt Booher suggested adding links to video clips to the out-by-out strip at the top, for key plays. So, now, if you hover over game out #16 (6th inning out #1) or game out #22 (8th inning out #1), you will see links to the MLB video clips for these outs. A new browser window opens in each case.
Try out the viz below, and think of a "Yes, and ..." that might apply to your work and data.
Recently, some colleagues of mine from Powerhouse Factories presented at The Ohio State University seminar on Big Data: Perils and Promises. Their presentation is available on the web at:
We have tweeted this a number of times, and it has received some visits.
But I got to thinking that there ought to be some other ways of socializing this content and making it stick with visitors or our customers. Wouldn't it be terrific if they could be reminded easily what our position is on this subject while consuming and interacting with analytics and visualization content we have developed for them. Here is where Tableau is quite useful.
At the recent Tableau European Customers Conference, there was discussion in the twitter stream (https://twitter.com/#!/search/realtime/%23tcceu12) about embedding web content inside Tableau deliverables. The specific question had to do with embedding Tableau views and workbooks into PowerPoint presos, and the other way around -- embedded PowerPoint presos into Tableau workbooks. Secondarily, what about other content, like web sites, .pdf documents, etc. All this is quite simple with Tableau using URL actions. Their knowledge base elucidates these examples: Linking to External Files. and Embedding Live Tableau Server Views into PowerPoint. With this guidance, embedding the SlideShare presentation in a Tableau workbook we have developed for a client should be a no-brainer.
And, it is. In the Tableau workbook below, our client wanted to see the location and number of direct mail prospects that might be available for a new store opening, and they wanted to interactively modify the criteria for the household demographics and distance from the proposed store. While we were at it, we took the opportunity to embed the SlideShare preso into the workbook, and delivered the whole thing via Tableau Public, since there was no proprietary data in the workbook. The views in this embedded workbook are accessible using the tabs at the top. If you click on the tab titled "Humanizing Big Data for Insights and Actions", you should see the same SlideShare presentation. This could very easily be a results presentation, or a video explaining how to use the rest of the workbook, or the latest Flickr photostream from the trip to grandma's house.
