Now Presenting! Presentations about the Presentation of Data, for Data Presenters and those that like to Present Data
Saturday, April 17, 2010
Wednesday, April 14, 2010
New Stephen Few blog post
Here is a great new blog post by Stephen Few talking about how to make the right decisions creating data visualizations rather than letting software make the decisions for us. He even references our patron saint, William Playfair!
Monday, April 12, 2010
Hey, This Chart Isn't Mathematically Accurate!
I made a quick purchase on a site today, and got one of those Bizrate.com survey requests. One of those, "Take this survey, you might win $25" offers (and I like free money).
After taking the 3-question survey, I got this summary of my responses. This is the actual screenshot of how my scores measured. Take a good look, and you might see what is wrong:
But at least I might win $25!
By the way, I'm also surprised a 7/10 got a satisfaction rating on the scientific smiley-face scale of
:)
Unless a 10/10 is a
:D
In which case, it's accurate!
After taking the 3-question survey, I got this summary of my responses. This is the actual screenshot of how my scores measured. Take a good look, and you might see what is wrong:
But at least I might win $25!
By the way, I'm also surprised a 7/10 got a satisfaction rating on the scientific smiley-face scale of
:)
Unless a 10/10 is a
:D
In which case, it's accurate!
Friday, April 9, 2010
Wednesday, April 7, 2010
Walmart Growth in Flash Data Visualization!
This is great:
Walmart growth since the 1970's
And how he did it!
We're all fans of mapping by zip code (and Tableau, one of our favorite data visualization tools, has that built in). Animating in Flash might not be possible right now. But a year-to-year view of something similar? I think that's a challenge one of us running this blog should tackle!!
Walmart growth since the 1970's
And how he did it!
We're all fans of mapping by zip code (and Tableau, one of our favorite data visualization tools, has that built in). Animating in Flash might not be possible right now. But a year-to-year view of something similar? I think that's a challenge one of us running this blog should tackle!!
Friday, April 2, 2010
Blog post about Stephen Few's blog
Here's a recent entry related to dashboard design on Stephen Few's blog:
Thursday, April 1, 2010
Dashboard Critique - Part One
Dan recently put up this post, from a company's Flash-based dashboard product. Take a minute if you haven't already at the visuals:
http://playfairgroup.blogspot.com/2010/04/i-recently-got-e-mail-for-flash-based.html
I started to dissect the issues in the comments, but realized this probably needed its own entry. And once I started going, I realized this might need several entries!
I hate to tear apart others' visualizations, but it is hard to look beyond the poor choices made in this dashboard - both in chart type and formatting. And it's not necessarily this company who is alone in making these types of mistakes. It seems to be a common issue in these one-size-fits-all template dashboards, where you may be able to throw a chart here or there, though the thought seems to be missing for what you should see, and how you should see it. And of course, the formatting. Egads, the formatting. But more on that as we go on.
So here we go....
**********************************************************
Let's start up on the upper left and lower left charts, the two column charts of this dashboard:
Not knowing much about this data, it seems that the choice of a column chart here was spot-on. When you are looking to compare different aspects of like groups (here Budget/Actual/Anticipated, and Actual/Anticipated, of these business segments). What ruins these? Mostly, the size of the chart itself. I don't think there should ever be a reason to have more than 50% of the space in your chart display be for data labeling - and this must be at least 80%. With South Dashboards, is Anticipated higher than Budget? Is the actual percent complete of North Dashboards higher or lower than anticipated? I think once they see the y-axis labels running into each other like this, it's time to stretch things out.
Fixes? Either shrink the size of the text (make 2 lines, or abbrev. depts.), or increase the size of the graph's space. You know, Excel chart manipulation 101.
(There also may be too much rounding of the column edges to make accurate readings, but I can't tell for sure within the half inch of space provided...)
**********************************************************
Let's finish off the left side with arguably the most obvious waste-of-space on this dashboard, the traffic light:
The purpose of this chart is understood - it is trying to communicate a simple point: this project is behind schedule. But even a gauge chart (which I will get to later) give more meaning, since you can at least judge how far you are behind a target. The sad, lonely "-13%" floating at the bottom of the graphic makes me yearn for so much more. Oh what a useful chart you could have been!
Now, I am a fan of traffic lights in data visualization. Their colors are universally meaningful - RED=BAD/STOP, YELLOW=OK/CAREFUL, GREEN=GOOD/GO! In fact, the colors are so meaningful on their own, that you really don't have to build a physical traffic light diagram on which to display them.
Not to mention, of course, it takes up such valuable real estate on this dashboard with an essentially meaningless image. (The waste of space is particularly sad given the content both above and below so desparately needing more room.)
So ignoring the possibility of charting the percentages, and simply using the traffic light approach, here is an example of how to make this a more efficient use of space, and how to show, generally, how to use the traffic light concept to track projects. (Which the title of this module, "General Project Trend", has me believe is the point.)
Yes, a bad mock-up using MS Paint, but you get the idea. Some projects are in danger, some are close, and others are on the right track. Quick. Easy.
**********************************************************
So that about covers the left side of the dashboard. Coming up new posts, I will dissect the other areas of the dashboard, and look at the build as a whole.
http://playfairgroup.blogspot.com/2010/04/i-recently-got-e-mail-for-flash-based.html
I started to dissect the issues in the comments, but realized this probably needed its own entry. And once I started going, I realized this might need several entries!
I hate to tear apart others' visualizations, but it is hard to look beyond the poor choices made in this dashboard - both in chart type and formatting. And it's not necessarily this company who is alone in making these types of mistakes. It seems to be a common issue in these one-size-fits-all template dashboards, where you may be able to throw a chart here or there, though the thought seems to be missing for what you should see, and how you should see it. And of course, the formatting. Egads, the formatting. But more on that as we go on.
So here we go....
**********************************************************
Let's start up on the upper left and lower left charts, the two column charts of this dashboard:
Not knowing much about this data, it seems that the choice of a column chart here was spot-on. When you are looking to compare different aspects of like groups (here Budget/Actual/Anticipated, and Actual/Anticipated, of these business segments). What ruins these? Mostly, the size of the chart itself. I don't think there should ever be a reason to have more than 50% of the space in your chart display be for data labeling - and this must be at least 80%. With South Dashboards, is Anticipated higher than Budget? Is the actual percent complete of North Dashboards higher or lower than anticipated? I think once they see the y-axis labels running into each other like this, it's time to stretch things out.
Fixes? Either shrink the size of the text (make 2 lines, or abbrev. depts.), or increase the size of the graph's space. You know, Excel chart manipulation 101.
(There also may be too much rounding of the column edges to make accurate readings, but I can't tell for sure within the half inch of space provided...)
**********************************************************
Let's finish off the left side with arguably the most obvious waste-of-space on this dashboard, the traffic light:
The purpose of this chart is understood - it is trying to communicate a simple point: this project is behind schedule. But even a gauge chart (which I will get to later) give more meaning, since you can at least judge how far you are behind a target. The sad, lonely "-13%" floating at the bottom of the graphic makes me yearn for so much more. Oh what a useful chart you could have been!
Now, I am a fan of traffic lights in data visualization. Their colors are universally meaningful - RED=BAD/STOP, YELLOW=OK/CAREFUL, GREEN=GOOD/GO! In fact, the colors are so meaningful on their own, that you really don't have to build a physical traffic light diagram on which to display them.
Not to mention, of course, it takes up such valuable real estate on this dashboard with an essentially meaningless image. (The waste of space is particularly sad given the content both above and below so desparately needing more room.)
So ignoring the possibility of charting the percentages, and simply using the traffic light approach, here is an example of how to make this a more efficient use of space, and how to show, generally, how to use the traffic light concept to track projects. (Which the title of this module, "General Project Trend", has me believe is the point.)
Yes, a bad mock-up using MS Paint, but you get the idea. Some projects are in danger, some are close, and others are on the right track. Quick. Easy.
**********************************************************
So that about covers the left side of the dashboard. Coming up new posts, I will dissect the other areas of the dashboard, and look at the build as a whole.
I recently got an e-mail for a flash-based dashboard company with a link to their demo. (See screen shot below). How much of this is useful information and how much of the dashboard is wasted on chart junk? Seems like this space could be used a lot more effectively, but is this what people are asking for on their dashboards?
Subscribe to:
Posts (Atom)