Perceptual Edge Discussion Forum

Confusing diagram from EIA

EAT — Wed, 18 Nov 2009 21:58:11 GMT

Hi All,

I recently went through all Few's articles associated with his diagram Examples with great interest. I feel that I've gained many insights as a result.

Just now I came across a diagram that I found confusing and difficult to decipher. So I thought I'd create an account and post it here to see what might happen with it:

http://www.eia.doe.gov/emeu/aer/pecss_diagram.html

Forum: Examples of bad graphs and dashboards

How to roll-up data but not lose detail?

mjoyce — Sat, 14 Nov 2009 03:25:29 GMT

Hi all,

I have been mulling over various way to report some data.
I'm responsible for various Information Technology service, through Asia.
The infrastructure span 11 countries which are grouped into three regions. Each country would have several sites.

We monitor service availability and strive for 100%.
We often get 100% too.

The data I want to present is a monthly report showing availability.
After some thought (and Stephen's Dashboard book) I realise I actually want to show unavailability, outages; to highlight issues.

The problem I'm having is I'm not sure how to display very small differences.

For instance a country with 5 sites, and one site has a 1 hour outage. 1 hour in a month is 0.13%.

Site1 Site2 Site3 Site4 Site5
99.87% 100.00% 100.00% 100.00% 100.00%

If the outage was a whole day:
Site1 Site2 Site3 Site4 Site5
96.77% 100.00% 100.00% 100.00% 100.00%

What chart will draw attention to the outage?
A column chart would I I alter the y scale, but I don;t like that as each month the scale will be different.

If I roll up the country using hi/lo/av, with only 1 hour outage :
Hi Lo Av
100.00% 99.87% 99.98%

Then if i roll that up again, with a bunch of countries, the outage kind of gets smoothed away.

It occurred to me that as I know the population of each site, I can calculate the impact in hours, and present both the rolled up averaged availability% and the summed impact hours.

I'm going to mock up a xls of data, for n countries with n site and simulate some outages.

Does anyone have any experience of this sort of data?

Regards

Matt

Forum: Graph design

Welcome

bpierce — Fri, 13 Nov 2009 00:59:35 GMT

Welcome to this discussion forum about Stephen Few's book Now You See It: Simple Visualization Techniques for Quantitative Analysis. The purpose of this forum is to give you an opportunity to make comments or to ask questions about the book.

-Bryan Pierce

Forum: The book "Now You See It"

Progress against a goal when lower is better

InfiniteJoy — Wed, 11 Nov 2009 21:55:53 GMT

How would you best visualize progress against a goal when lower is better?

Most bullet graphs or bar graphs with a goal line indicate to a user that you want to "meet or exceed" the goal.

In this case, I want to be "at or below" the goal, so I suppose goal really means upper limit.

Here is the general feel of my dashboard. The metric on the left is a "meet or exceed" metric. The metric on the right is "at or below" and lower is better.

Forum: Visual data analysis techniques

Motorcycle Accident Statistics

RCHaynes — Tue, 10 Nov 2009 17:50:19 GMT

As an avid motorcyclist I am always searching for more data and better representations so that I can explain why the mantra 'All The Gear All The Time' is a wise approach.
And now we get this posted by a member: http://www.rideamoto.com//uploads/images/0000/0802/MOTO-DANGERS-R2.png?1257663823 published by the NHTSA.
The poster has determined from this that 3-6 pm is the most dangerous time to ride, and that a helmet is only useful a bit over 1/2 of the time.

So I thought I'd put this up so that folks could shred the design flaws.

Forum: Examples of bad graphs and dashboards

Clearquest

wd — Thur, 05 Nov 2009 16:40:01 GMT

It's a software development defect tracking tool that found its way into our company to help track a major project. While its intent may be clear, the output is a tad foggy ...

I was able to show the people involved an alternative display that does a better job, but Clearquest defaults aren't all that adjustable. Might be a great defect tracker, but it is not a defect reporter!

Forum: Examples of bad graphs and dashboards

Yet another review request

PaulM — Sun, 01 Nov 2009 09:09:41 GMT

The attached dashboard shows the at risk sectors for the Road Safety Authority. When the report loads it defaults to showing year to date, the user can then use the menu to choose a different end date. As this is targeted to Israeli users, everything is in Hebrew and the date menu is on the right.

The labels on the right are the names for each metric. The underlined labels are drillthroughs that open related reports. Eventually they will all have related reports.

The numbers, going from right to left are: No Greater Than (can't use the word target), Selected Period, Difference of Target and Selected Period, Previous Period.

The line graphs specs are: Black is selected period, gray is previous. The chart always shows a 12 month period. Right now it shows December-November. So the gray December is actually for December 2007.

As Cognos does not yet include bullet graphs I'm using Google charts. The limit line is always two thirds in. The dark gray bar shows the value for the previous year. Any time a metric exceeds the limit a red dot appears next to it.

It took a lot convincing to get the users to agree to "ugly" dashboards, but I think they see the difference. This gives me a little more maneuverability when it comes to designing effective reports. So please feel free to rip this to shreds.

Thanks,
-Paul

Forum: Dashboard design

Homeland Security Cybersecurity Center

jmunoz — Sat, 31 Oct 2009 02:45:34 GMT

Today the NYTimes showed a photo in their "Pictures of the Day" section of Homeland Security's Cybersecurity and Communications Integration Center. The dashboard in the background jumped out for all the wrong reasons. There must be a 3D mandate for all Homeland Security graphs. Someone, please rush the Department of Homeland Security copies of all of Stephen Few's books, especially Information Dashboard Design!

John C. Munoz
http://www.bzintelguru.com

Forum: Examples of bad graphs and dashboards

Fundamental Differences in Analytical Tools

bpierce — Thur, 29 Oct 2009 16:37:10 GMT

In his September/October 2009 article, titled "Fundamental Differences in Analytical Tools: Exploratory, Custom, or Customizable," Stephen explained that there are three basic categories of analytical tools, each intended for a specific audience and suited for specific types of analytical tasks. Do you agree? Is there something you'd like to add?

-Bryan

Forum: Articles by Stephen Few

Please Critique my Dashboard

smbradbeer — Thur, 29 Oct 2009 12:27:42 GMT

Hi, I have done a simple 'dashboard' showing the results of an annual survey and I was hoping you'd provide some constructive criticism.

I used MooD software, which has poor charting functionality, so I can't change the colours and don't have much control over labels in the line graph.

The little blue triangles indicate where users can drill down (to historic data)and I am not able to remove them.

I did consider only showing RAG icons for the red (and amber?) measures but I'm not sure how that would go down (they like to see green!).

Anyway, please be gentle as I am a sensitive soul!

Thanks,

Sarah

Forum: Dashboard design

Info Vis history on Flickr

jmunoz — Tue, 27 Oct 2009 15:19:57 GMT

This morning I stumbled across a tweet about a cool info vis book. I just had to check it out.

The book, titled "Graphic Presentation" and written by Willard Brinton is 70 years old. I was surprised to find so much good stuff in the book, but maybe I shouldn't have been. In the days before computers and Excel, when making a chart took a serious investment in time, the majority of chart makers really thought about their craft. Contrast that to today, when making a chart in Excel consists of hitting the F11 button, so easy, even a caveman can do it.

One of my favorite images from the book is below. At first, I couldn't figure out the order of the regions on the chart on page 143, but then it hit me. Can you figure it out? In addition, the use of small multiples on the bar charts on 142 makes it easy to compare across the categories.

The Flickr stream, belonging to Prof. Michael Stoll, can be viewed here and the entire book can be downloaded (low res, sorry), here.

Enjoy.

John C. Munoz
http://www.bzintelguru.com

Forum: Visual narrative

Bowling Alone chart

gmklass — Wed, 21 Oct 2009 13:35:17 GMT

From Robert Putnam's Bowling Alone:

In addition to the 3-D, the chart does not display meaningful data: the most important data (changes over time) are actually represented by the arrows on the bars, not by the height of the bars).

The chart is not self-explanatory .

Attached a revised chart based on what data could be gleaned from the text.

--gary klass

Forum: Examples of bad graphs and dashboards

Bullet Graph Generator available - Open Source

ChrisGerrard — Wed, 21 Oct 2009 02:47:55 GMT

I'm happy to announce that The Bullet Grapherator is ready for public exposure.

An Open Source project creating Bullet Graphs in PNG, JPEG, and SVG formats, it's online at: http://kenai.com/projects/bulletgrapherator.

See GettingStarted for instructions on how to get it. (The zip file is a bit too big for attaching here, I think)

In its current form there's a Designer within which you can configure your Bullet Graphs to a reasonable degree, with much more configuration flexibility in store. Once designed, individual Bullet Graphs can be stored as templates, which can then be loaded and supplied with the variable business data to be displayed. There's no well-defined boundary as to what this will be, but it's a good bet that it'll include the upper and lower limits, the performance value, the comparison value, and less frequently the qualitative ranges. All of these can be provided programmatically when rendering updated BGs with new data for inclusion in your dashboards.

The Grapherator started as a way to automate the creation of SVG Bullet Graphs. As things got more involved incorporating PNG and JPEG format generation seemed only right.

The Grapherator is written in Java, and so is most naturally used in Java environments. Which means that it can be slotted into Java-based web apps pretty easily, although there's work to be done to expose the programmatic API better - it's still pretty ad hoc.

My hope is that The Grapherator will prove to be useful, and that there will be enough interest in it to make it worthwhile to continue to working on it.

So, please take a look and let me know what you think.

Be kind, but honest.

Pictures attached.

Forum: Visual data analysis software

We are what we eat

jmunoz — Tue, 13 Oct 2009 12:54:43 GMT

Here's are some tastefully designed cupcakes, with the exception of the two pie cakes and the heavy tick marks on the axes.

Original photo here.

Forum: Visual narrative

To Radar Graph or not to Radar Graph...?

quetzalc0atl — Wed, 07 Oct 2009 10:59:39 GMT

Hello,

I'm currently working on neighbourhood (N'hood) area profiles for our city - the city is made up of 43 neighbourhoods.

We've created different thematic baskets which are made up of various KPIs. Each of these thematic baskets has an overall score - the closer this score is to 100 the more deprived the area.

One of things that we are looking at in these profiles is to do a comparison between a N'hood, the City and the best N'hood (i.e. the least deprived n'hood overall). The idea is that people who live in these neighbourhoods can see how poorly their neighbourhood is doing and then challenge their local politicians to improve theses areas.

It had been suggested that using a radar graph would easily allow members of the public to compare their N'hood with the city average and the best N'hood. However this has come under some criticism from several senior analysts as they feel that it would confuse members of the public too much.

Having reviewed some of the literature on the area and looked at the suggestions from Mr Few in the Examples area of the website. I've come up with some alternatives and was wondering whether I could get your opinions on what is most suitable.

I'm a relative newbie - but I feel that the radar graph although not as clear as some of the others still provides the best overall picture of things in terms of what we are trying to do.

Thanks,

Awais

Forum: Graph design