Back in January, I published a post on how to create a calendar-based view in Tableau. At the time, I mentioned that it had created some buzz within the halls at Tableau-central and that, in part, lead to an invitation from Christian Chabot, CEO, to participate in a new project they were working on.

Well, here is your first peek at that project – Tableau Online!

Essentially, it will allow anyone with Tableau Desktop to publish their workbook online and embed the visualization in their website.

Tableau Online is currently in private beta, but you can sign up for updates here, or click on the logo below the visualization.

Before we get to the visualization itself, here’s what I’ve learned about the project from Christian and Ellie Fields – Director of Product Marketing.

Christian says that the vision for Tableau Online is for it to be “the YouTube of information visualization…” which I interpret as the platform of choice for publishing rich, interactive visualizations online.
Additionally, during a recent discussion with Ellie Fields, I uncovered some of the more nitty-gritty details…

1. It currently only works with local data – basically you need an extract from any source locally on your machine in order to use the visualization with Tableau Online (in all likelihood you will upload a packaged workbook)
2. There have been discussions around an API so that data might be transmitted ‘over-the-wire’, but there are no specific plans for getting that done
3. This is a very early look, the hardware is still undergoing performance testing so be patient if you get an error with the visualization or it takes a while to load; and hey, leave a comment describing what happened!

What follows is a proof of concept based on the calendar view I had developed.

This calendar specifically, and Tableau Online generally is built on the new version 5 platform (there have been some examples creeping out on the Tableau blog). I’ve been working with the version 5 beta so let me tell you about the features from 5 that are important here. Sheetlinks and highlights have been repackaged – and improved – to a new function called ‘Actions’. In this visualization, the calendar filter for year is also applied to the small multiples sparklines below the calendar. This is now very easy to do compared to how sheet links work in version 4. Also, if you click on any particular day within the calendar, that date is highlighted on the small multiples below – enabling this is as easy as hitting a switch (which is basically what you do).

Many improvements and new options have been added to quick filters – but, so far, the biggest improvement IMHO is the addition of the search feature! (Although I wish that the Tableau team had built in direct feedback for when a search fails)

Here are some of the things you can do in this proof of concept.

1. Filter on Year for the whole visualization
2. Select a particular month for the calendar view
3. Highlight any given day in both the calendar and the small multiples by selecting a day in the calendar (you can also select multiple days)
4. Highlight a day in the small multiples only by clicking on a plotted point (or drag select multiple days)
5. Highlight a particular month by selecting the month name in the small multiples column header
6. Lowlight inactive days in the calendar for a particular product category by selecting the category name in the small multiples row header (you can select multiple categories by selecting one category and then either SHIFT or CTRL clicking another category)

The Tableau team and I have discussed many changes to this visualization, but we weren’t able to get them done in time for this post as each change currently requires extensive testing on the new platform, but those changes should be coming soon. Discussed changes include changing the color encoding to profit instead of total sales, and doing a daily sparkline only for the currently selected month and then bar charts aggregated by month for the rest of the months in the currently selected year.

What else would you change/remove/add to this visualization?

Finally, there will be a smattering of sneak peeks going up across the infoviz community so keep your eye on out here and the Tableau blog as those new visualizations go up!

Science News has a short little article on Florence Nightengale’s impact on data visualization.

From the article:

…As impressive as her statistics were, Nightingale worried that Queen Victoria’s eyes would glaze over as she scanned the tables. So Nightingale devised clever ways of presenting the information in charts. Statistics had been presented using graphics only a few times previously, and perhaps never to persuade people of the need for social change. In doing so, she ignored the express advice of her mentor, Farr. “You complain that your report would be dry,” he wrote to her. “The dryer [sic] the better. Statistics should be the dryest [sic] of all reading.”…

They have what I’m sure is a fun little interactive graphic of her “coxcomb” info-graphics but I’m not sure it’s helpful.

What do you think?

Heather Dougherty, Director of Research over at Hitwise recently published a post about the geographic sources of traffic to the U.S. President Candidate (Barack Obama & John McCain) websites compared to how likely a particular state is to vote for that candidate (as published by the NY Times).

I can see what Heather was attempting to do with the color choices in her table – trying to visually relate each row in her table to the geographic map produced by the NY Times. However, there are a couple of problems with that. First of all, the seemingly bold choices of the colors on the US Map do serve an important purpose – they make each state’s affiliation immediately clear. Secondly, the colors are carefully paired. The pure dark blue of Obama and pure dark red of McCain have the same intensity so that our eyes aren’t tricked into thinking there’s more blue or red than there actually is or that one is more important than the other. The secondary colors appear to be less intense than the pure colors but are also matched in intensity to each other. The yellow of the ‘Tossup’ states are actually visually striking – on purpose I’m sure.

However, when those colors are poorly reproduced in a table what you end up with is a hot mess – visually.

And is it just me or is the choice of dark blue for the column headers awkward? It tends to make me think that the table itself is leaning towards Obama when that’s probably not the intent. As you can see by comparing the table to the map, there was an attempt to relate the states in the map to the rows in the tables by giving them -similar- background colors, however the matching is poor and the table ends up being difficult to read.

On an analytical note, it sure would be nice if the tables provided total % of traffic to each candidate’s site from the top 10 states (easy enough to calculate) as well as the total “Representation Index against online pop” (whoa, that’s a mouthful, couldn’t we just call it Online Index like other audience measurement services and then provide a definition if we feel it’s required?)

Below, are the tables, as I have redesigned them:

In the screen-shot above I included the original NY Times map as reference. As you can see, I’ve removed all the color matching within the rows themselves. Instead, by taking a look at how the colors are applied to the map:

We can see that the groups are easily referenced as a numerical score (-2 to +2):

Not only will this value assignment allow us to clean up the table, but it creates the opportunity to perform some simple operations on the scoring – more on that in a bit. So, as you can see while still containing all the information of the original table, we have cleaned it up quite a bit and hopefully made it more readable. Now, because it’s not a bad idea to make sure that the table data is still easily relatable to all the information in the NY Times map, I went ahead and created legends for the state-scoring column. Oh, and I did take the extra five minutes needed to pop the map into GIMP so I could use the eyedropper tool to get the exact RGB values for each color (a good practice – your eyes can and will fool you).

For easy comparison, here are the two tables:

Original Table by Hitwise

Our Cleaned-Up Table

Now getting back to wishing that Hitwise had provided totals for each of their columns (Traffic Share and Online Index) and what the numerical scoring does for us. Note that we’ve included a “TTL/AVG/TTL” row at the bottom of the new table. The Online Index is averaged because we don’t have access to the raw data needed to compute the actual Online Index for those top 10 sites.

But in any case, we can see that the Obama site traffic dominated by states that are pro-Obama (+5) while the McCain site would appear to be dominated (-7) by pro-Obama traffic as well. That’s interesting and leads to all kinds of questions about the demographic and psychographic makeup of those visitors. Also, the average Online Index for the Obama site is much higher than that for the McCain site. I’m not sure if this indicates that Obama’s traffic is highly concentrated or that McCain’s is more distributed but it certainly begs to be explored.

In any case, this reminds me that actual design standards in reports and analysis are important – why do you think Jupiter Research and Forrester seem to have templated charts and tables (oh hey, aren’t they the same company now?) Despite the red, Jupiter’s report and infovis templates are highly readable – the same for Forrester’s green (will it all turn into christmas trees now that they are one?)

If you haven’t already you should sit down with your analytics team and hash out the following:

1. Standard Color Palette – be consistent in your application of color

2. Standard Elements for Charts

3. Standard Sized for Charts

4. Standard Application of Chart Types to Dimension Types

5. Standards for Fonts (type-face, size etc)

6. And yes, Standards for table design

By using the same conventions over and over, you not only save yourself a lot of busy work, but you encourage your audience by re-enforcing usage through design, it makes it easier for them to grok how to use something if they see it over and over again.