Perceptual Edge Discussion Forum

McDonald's incell bar charts for nutritional stats

jmunoz — Thur, 02 Jul 2009 14:58:17 GMT

As an info viz junkie I was excited to see McDonalds using incell bar charts on their food wrappers. I've attached an image of one off of a cheeseburger wrapper.

However, as a Stephen Few follower, I know that a key ingredient to high quality info viz is simplicity--it's got to be easy to understand. To that end, what in the world does the dashed vertical line spanning calories through sodium represent? It looks like it's set to 33%, but why? Is that some magic number in the world of nutrition? Is 33% "crossing the line" for a single item on their menu? There's ample room for an explanation of the reference line on the wrapper, but none is provided.

McDonalds has missed the mark on a few other things on their graphics. The icons to the right of each item are meaningless and the heavy border surrounding each item detracts from the data, but the reference line to me is the real head scratcher.

McDonalds has a great opportunity to educate their customers about how the calories/sodium/protein/fat/carbs they're eating right now fit into their overall daily intake of calories/sodium/protein/fat/carbs. I get the sense that the incell bar charts are an attempt to simplify this information, which is great. But their attempt comes up short because they didn't explain the reference line.

Lastly, has anyone else seen these graphs on their wrappers at McDonalds? I'm curious if this is a local test. I live near Philadelphia, PA.

Forum: Graph design

How far to push simplicity?

thomasimmich — Thur, 25 Jun 2009 17:06:33 GMT

Just read your book and very much liked most of your statements. In particular, I share with you the philosophy of keeping things simple and to strictly avoid arbitrary design decisions.

However, with a few of your principles I disagree, especially those that - in a quite dogmatic way - abandon visual elements that might not add any "value" to a dashboard. I think what's "valueable" is pretty discussable in general and should be communicated as such. E.g. I personally prefer adding properly muted) grid lines to all my charts - though they are indeed not obligatory to effectively communicate data in most cases, I prefer using them as they make it easier and more relaxing to associate a data point e.g. in a line chart with a corresponding accurate value - especially if the data  
point is far away from the chart's axes. It is hard to tell if users really never need to know a more accurate value to a chart's data point so I add grid lines as a common means not as an exception.

But this is just an example. What I'd like to state is that variety in design and very few times breaking out of principles - if applied with care and supported by proper user testing - does *not* necessarily undermine the good design of a dashboard (or GUIs in general). In case you share the same opinion, this is a message I would have liked to come through more clearly in your argumentation. In case not, I would  
have liked to read a more detailed reasoning.

Forum: The book "Information Dashboard Design"

Spinning Pies - Oh My!

grasshopper — Wed, 24 Jun 2009 13:11:15 GMT

Recently on a dashboarding website, someone (anonymous) posted the following comments, in regard to dashboarding on iPhones...

"It looks like the team at MeLLmo is light years ahead with the RoamBi product. http://www.roambi.com "

I had a look at the demo/video on the website. I guess it's quite interactive, and makes use of many of the iPhone GUI features, but as far as their end-result, I was quite disappointed. Here are some of the things that jumped out at me...

Black Backgrounds throughout.
Gradient fill in their bar charts.
When you scroll up/down through the bar chart, I don't think you can see the scale anymore.
And, last (but not least!) ... Spinning Pie Charts!

Per the spinning pie charts, it was kind of like "Wheel of Fortune" ... you grab the edge of the pie and "spin" it, and it shows the value of whatever slice the pointer is pointing to - I really expected to hear the wheel-of-foutune click/click/click sound effect! ;) And the pie has a bit of a 3d curve, with candy-sheen, look to it, if I'm not mistaken.

Forum: Examples of bad graphs and dashboards

Dashboard/Report Draft for Critique

nick3412 — Thur, 18 Jun 2009 12:23:13 GMT

So I am a student entering my senior year in college, and I have been interning at a business intelligence company for the past three years.

This is a rough dashboard/report I quickly threw together today. It is a financial/sales performance report for a fictional pizza place. I am toying with the idea of distributing a report like this locally myself, independent from my internship.

Any critique(s) would be greatly appreciated on the design/content of this report.

Thanks in advance.

Forum: Dashboard design

What is "the" colour circle?

Derek_C — Wed, 17 Jun 2009 23:01:33 GMT

I was reading about the classification of of colour schemes as complementary, triadic, tetradic etc., when I suddenly realized that the language only makes sense if there is one true colour circle! I already knew that there were colour circles where the colours appear at different angles, stretching some parts of the circle and compressing others, and I was okay with that. But now it seems to me that it can't be okay to have different wheels, if the aesthetics of "primary", complementary, and tertiary colours, and so on, are to be laid down so strongly.

As we screen-readers know, computers typically use a red-green-blue triad as the basis of their colour circle, and printers use the equivalent yellow-cyan-magenta complements. But colour designers, as in this article by Maureen Stone, typically use a red-yellow-blue triad, with green-purple-orange complements, to create a red-orange-yellow-green-blue-purple circle. There's no way around it. If orange "is" blue's complement, then yellow can't be, and if green "is" red's complement, then it can't be cyan. If a triadic scheme consists of red, yellow and blue, then it isn't possible to argue that red, green and blue are a triadic colour scheme.

So, is there evidence that one particular colour circle is the "true" circle for aesthetic purposes, and if so, why? I notice that the red-yellow-blue system beloved of painters and other arty folk doesn't match the Natural Color System (NCS), which is supposed to be based on physiology, and has a colour circle based on a tetrad of red-yellow-green-blue.

Forum: Graphicacy

Cartographic Malpractice

grasshopper — Tue, 16 Jun 2009 19:16:42 GMT

I liked this article, as it's a type of chart I like, and I also agree with
Stephen's observations about good & bad practices :)

In particular, I agree that with these types of charts the days (or whatever)
must be kept discrete (no color blending across days in a contour fashion).
And yes, a legend should be included (duh!)

As a designer of such charts, I frequently get requests for the gradient
contour shading, and I "Just Say No" :)

Here are a few examples of this kind of chart I've done in the past
(they're not perfect, but I think they follow the basic rules of "good"
data visualization ... and common sense :)

http://robslink.com/SAS/democd1/calgrid.htm

http://robslink.com/SAS/democd14/allergy.htm

http://robslink.com/SAS/democd22/trend.htm

http://robslink.com/SAS/democd24/beach.htm

Forum: Articles by Stephen Few

Ranking the Gestalt Principles

Derek_C — Sat, 13 Jun 2009 11:57:54 GMT

In Chapter 6 you say the Principle of Connection usually trumps the principles of Proximity and similarity, but not Enclosure. Is there a widely accepted ranking of these principles, in the spirit of Cleveland's Hierarchy? I made a start on such a ranking below, but I'd rather not conduct a personal program of experiments if the literature already contains a well-regarded answer.

1Enclosure
2Connection
3Proximity
3Similarity
?Closure
?Continuity

Forum: The book "Show Me the Numbers"

Chart vs. diagram

NoFisch — Wed, 27 May 2009 23:15:10 GMT

I would like to clarify two fundamental terms frequently used in InfoVis: chart and diagram.

Chart
In The Grammar of Graphics Wilkinson states that charts are the equivalent to words in linguistics. He says that perceivable graphs created by grammatical rules are called graphics (p1). He further says that We often call graphics charts and that he shuns chart typologies because charts are usually instances of much more general objects (p2)
Further he defines a graph being a set of points, and A mathematical graph cannot be seen. It is an abstraction. A graphic, however, is a physical representation of a graph. This representation is accomplished by realizing graphs with aesthetic attributes such as size or color. (p6)
I just read on http://en.wikipedia.org/wiki/Chart that a chart is a visual representation of data, in which the data are represented by symbols such as bars in a bar chart or lines in a line chart. A chart can represent tabular numeric data, functions or some kinds of qualitative structures.
A data chart is a type of diagram or graph, that organizes and represents a set of numerical or qualitative data, including maps and some forms of ordered tables.

Diagram
On http://en.wikipedia.org/wiki/Diagram diagrams are defined as pictorial, yet abstract, representations of information, which include maps, line graphs, bar charts, engineering blueprints, and architects' sketches. In the same article a citation by Lowe (1993) only includes area diagrams, flow charts, idea diagrams, and PERT charts in the diagram category.
The definition given is: 1. A collective term for any visual information device, like the term "illustration" often used as a representative term, to stand for the whole class of technical genres, including graphs and tables. 2. The specific class of visual display, that show qualitative data with shapes that are connected by lines, arrows, or other visual links.
It says further: Charts can contain both quantitative and qualitative information.
On the same page I found: Diagrams, according to Lee E. Brasseur (2003), are essentially drawings with text. They consist of basic lines and shapes that convey an idea.

Should I conclude that the main distinction between both terms is that a chart displays quantitative and qualitative data and a graph merely shows qualitative data?

Norman

Forum: New topic proposals

Budget bullets graph idea

seamonk — Wed, 20 May 2009 17:11:58 GMT

I'm working on a dashboard idea for the continuing education arm of the University of Washington. The idea is to produce a report for each of the Assistant Directors showing a list of their current programs, and how they are doing financially. The concept I've come up with is two matched bullet graphs, and I'd appreciate any advice this sage group can offer.

I've attached two examples. One is a large scale that would only be used for demonstration purposes to explain the mini graph. I hope things are fairly obvious - otherwise I haven't done my job well! I've converted all the dollar values to a percentage of the "capacity" of the program. The capacity is the total revenue possible if every class were filled to capacity with students paying the full course rate. The expenses are shown in the colored background, with the revenue as the dark "bullet path"

The upper bullet shows the budgeted amount, while the lower bullet shows the actuals through the current month. The reason the expenses have been broken out as they have is for planning purposes. As a non-profit, the revenue should closely match the expenses. However, there is a no-go decision that we make just before we offer classes - if enrollments are lower than expected we may still want to offer the class, so long as revenue covers the direct costs ("Faculty&Staff" and "other").

The reason the values have been converted to a percentage of capacity is so that a row of Programs will have graphs in a similar scale. In practice, the row would include values and ratios much as a stock does - standard dashboard fair.

The complications I face are:
1) needing two bullet graphs, over-under, per line
2) needing 5 (or even 6) background areas, versus a more typical 3 (in range, above range, below range)
3) we offer hundreds of programs, and thousands of individual courses that we might want to include in the reports, so easy automation is a must.

So, any advice? Especially, does anyone know if I could do this with XLCubed Microcharts?

Forum: Examples of bad graphs and dashboards

Appropriate scales on sparklines

Embino — Mon, 18 May 2009 02:46:54 GMT

Hi everyone

I'm pulling together report in which I am very keen to use sparklines. These are quite new for the organisation, so I'm keen to make them as meaningful as possible - the issue that I have is how best to scale the axes for them. I want the scale to be narrow enough so that the patterns of change are apparent, but as we know when you start messing around with scales you can mislead by making the amount of change appear greater than it is. Changing the scale on some of the sparklines I am working with makes them appear to have a similar rate of change, whereas one of them could have changed by 0.5%, and the other by 50%.

I'm not sure if there are any 'rules' to scaling sparklines, but I don't want to be accused of misleading my audience, and therefore have this very useful reporting technique rejected.

Thanks in advance of advice!

Emma

Forum: Graph design

Bullet graph extension

Mac — Fri, 15 May 2009 11:23:07 GMT

Hi,
The use of bullet graph in order to compare a current measure and estimate projection to a future target is very powerful.

Sometimes my internal users ask to me to drill down the summary information across the main company dimension (Brand, Channel, Product Category, etc).
Instead of creating a sequential set of rows for each domain value of the dimension (n-bullet graph) I wonder if could be a good idea to extend the bullet graph concept to a radial graph (I enclosed an example as attachment : RadialBG.png).

Perhaps the interpretation of the graph is not immediate, but the idea of comparing a sort of stain with a perfect circle seems seems good to me!
I do not know if this type of chart already exists, otherwise what do you think about?

Thank you.
Paolo

Forum: Dashboard design

vizbybis2 - bis2 Product Review

Galt — Tue, 12 May 2009 18:47:41 GMT

Steven:

I just came across a new data visualization product called vizbybis2 made by a company called bis2 (bis-squared). Their website states that they offer "Super Graphics" to visually display thousands of variables in one view that allows the human eye to see highlights, patterns and relationships amongst all these variables. I noticed that you've done some consulting with this company and wondered if you'd be willing to classify/compare this product to others such as SAS, JMP, Spotfire and Tableau.

On the surface, the product looks like it would require more training to get a handle on interpreting their Super Graphic visual designs. I'd be curious if anyone here in the discussion area has demoed or purchased this product. Here is the website for those who are curious:

http://www.bis2.net/

Thanks in advance for any thoughts and feedback.

-Galt (aka "Joe")

Forum: Visual data analysis software

Heat Map usefully applied

pzajkowski — Thur, 07 May 2009 15:11:53 GMT

I tend to struggle with reading heat maps, especially tree maps that render multiple boxes of various sizes nested within one-another, all with different levels of saturation. They just don't do it for me.

Recently, however, I came across a blog wherein a heat map was overlaid on top of a storage warehouse with the goal of optimizes the location of items within the warehouse based on frequency of retrieval. The goal of the warehouse analysis was to reduce unnecessary foot traffic to items with higher frequency retrieval; items rarely retrieved should be located farther away from the warehouse entrance.

The use of heat map for this analysis makes sense to me, perhaps in part to the literal representation of the warehouse itself, rather than the abstract construction of some heatmaps that use nested boxes. Perhaps a tweek to this approach might be to use a gray-scale approach, rather than color: i.e., black = highest retrieval frequency, lightgray= least frequency.

Anyway, here are two graphics from the blog that illustrate an optimized warehouse vs the actual heat map of the warehouse.

Forum: Visual data analysis techniques

Dashboard for iPhone

grasshopper — Wed, 06 May 2009 17:38:39 GMT

I had seen this mock-up of a dashboard for an iPhone on the Dashboard Spy website a while back ...



And I decided to try to create the real-thing, and see if it was viewable/usable on an iPhone. I was pretty happy with the results, so I thought I'd share:

If you're interested in the SAS/Graph code I used to create it, that can be found here:

   http://robslink.com/SAS/democd38/iPhone_dashboard_info.htm

And here's a link to a straight webpage image of the dashboard (rather than a photograph), if you'd like a closer/better look at the details:

   http://robslink.com/SAS/democd38/iPhone_dashboard.htm

Forum: Dashboard design

A Bad EarthDay Chart

grasshopper — Wed, 22 Apr 2009 18:01:14 GMT

In celebration of EarthDay, I thought I'd post up a bad EarthDay graph I happened across...

Nothing new/uniquely bad - just the classic 3d bar chart, where you can't compare the heights of the bars, because of the 3d-tilt, perspective, etc - for example, in this chart the "South Lake Waste" bar looks shorter than the "North Lake Waste" bar, even though the value is higher.

Even when people have the best of intentions, if they create charts that misrepresent the data, they could do more harm than good - making their audience wary of *all* the numbers & graphs that the whole organization or movement publishes...

http://ndep.nv.gov/recycl/earth/earth05_chart1.gif

Forum: Examples of bad graphs and dashboards