Neat use of Tufte's sparklines: Airline pet incidents

22 Dec 2008 - 1:15pm
5 years ago
11 replies
2368 reads
Elise Edson
2007

Hi IxDAers... just thought I'd share this neat example of sparklines!
I was looking for pet travel information across the various airlines,
and I was surprised/delighted to see this:

http://www.petflight.com/pet-incidents/airlines

How have you integrated sparklines into your designs?

Elise

Comments

22 Dec 2008 - 1:38pm
Alex ONeal
2008

Excellent! I'm always on the hunt for live examples of good information
presentation. Thanks for sharing :-)

I use sparklines for presentations on metrics-based recommendations, site
SEO reporting, etc. The response has been very positive every time, and the
biggest compliment of all -- others have begun to do the same :-)

Alex

--
The best time to plant a tree is twenty years ago. The next best time is
now.

On Mon, Dec 22, 2008 at 10:15 AM, Elise Edson <elise.edson at gmail.com> wrote:

> Hi IxDAers... just thought I'd share this neat example of sparklines!
> I was looking for pet travel information across the various airlines,
> and I was surprised/delighted to see this:
>
> http://www.petflight.com/pet-incidents/airlines
>
> How have you integrated sparklines into your designs?
>
>

22 Dec 2008 - 1:40pm
Jackson Fox
2006

Looking at the Pet Flight site, I wonder if the lack of scale/context
is problematic. I know I was a bit shocked at all of the peaks in the
graphs when I saw them. Not knowing the scale of either axis leaves me
to make assumptions about the data. In this case, I would assume
American Airlines is going to kill my cat.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=36598

22 Dec 2008 - 1:56pm
SemanticWill
2007

Agreed, context and scale are important, and breaking out lost, injured,
killed as three separate information vectors by airline, overtime without
connecting the information points would have increased information density
while reducing chart junk.
~ will

"Where you innovate, how you innovate,
and what you innovate are design problems"

---------------------------------------------------------------------------------------------
Will Evans | User Experience Architect
tel: +1.617.281.1281 | will at semanticfoundry.com
aim: semanticwill
gtalk: semanticwill
twitter: semanticwill
skype: semanticwill
---------------------------------------------------------------------------------------------

On Mon, Dec 22, 2008 at 1:40 PM, Jackson Fox <jacksonfox at gmail.com> wrote:

> Looking at the Pet Flight site, I wonder if the lack of scale/context
> is problematic. I know I was a bit shocked at all of the peaks in the
> graphs when I saw them. Not knowing the scale of either axis leaves me
> to make assumptions about the data. In this case, I would assume
> American Airlines is going to kill my cat.
>
>
> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
> Posted from the new ixda.org
> http://www.ixda.org/discuss?post=36598
>
>
> ________________________________________________________________
> Welcome to the Interaction Design Association (IxDA)!
> To post to this list ....... discuss at ixda.org
> Unsubscribe ................ http://www.ixda.org/unsubscribe
> List Guidelines ............ http://www.ixda.org/guidelines
> List Help .................. http://www.ixda.org/help
>

22 Dec 2008 - 2:14pm
mark ahlenius
2008

Hi,

appreciate the reminder of this application. I attended Tufte's
presentation in Chicago probably about 10 years ago. It was one of the
best presentations/seminars I have ever attended. I bought all his
books and really enjoy them.

I am not sure, but "sparklines" seems like it might have stemmed from
his work with "Multiples in Space and Time" which is a chapter in his
book Visual Explanations. In this chapter he has a graphical summary
of a patient status (pg 111). What is really fascinating about this
graphic is how they time align specific dosages, blood levels, chemical
levels, etc. This page is a collection of small graphs for each of
these items. Thus as Tufte explained it, instead of a group of doctors
and medical professionals trying to review a patient's status via a 1"
thick stack of charts and test results, they have a 1-sheet page which
can give them an instant summary of the status. This approach allows
for mind-based data mining to see patterns and trends, etc. And it of
course can be applied to many other areas. If you have not looked at
this, I'd recommend it. Its a well source of ideas for new applications.

Sparklines reminded me of this talk and it also reminded me of
earthquake seismic measurement data.

One caution I would add to sites like petflight.com when reporting this
type of data would be to normalize it based on number of pet flights vs.
injuries. It wasn't clear to me that it was done. Airlines like
American and United will most likely have far more pets flying on them
vs. smaller airlines. We all know that you have to be very careful when
you present data, I see abuses in the news print all them time on how
they report a story with stats. But perhaps I missed that and its
already in the data presentation.

best,

'mark

Alexandra O'Neal wrote:
> Excellent! I'm always on the hunt for live examples of good information
> presentation. Thanks for sharing :-)
>
> I use sparklines for presentations on metrics-based recommendations, site
> SEO reporting, etc. The response has been very positive every time, and the
> biggest compliment of all -- others have begun to do the same :-)
>
> Alex
>
> --
> The best time to plant a tree is twenty years ago. The next best time is
> now.
>
>
> On Mon, Dec 22, 2008 at 10:15 AM, Elise Edson <elise.edson at gmail.com> wrote:
>
>
>> Hi IxDAers... just thought I'd share this neat example of sparklines!
>> I was looking for pet travel information across the various airlines,
>> and I was surprised/delighted to see this:
>>
>> http://www.petflight.com/pet-incidents/airlines
>>
>> How have you integrated sparklines into your designs?
>>
>>
>>
> ________________________________________________________________
> Welcome to the Interaction Design Association (IxDA)!
> To post to this list ....... discuss at ixda.org
> Unsubscribe ................ http://www.ixda.org/unsubscribe
> List Guidelines ............ http://www.ixda.org/guidelines
> List Help .................. http://www.ixda.org/help
>
>

22 Dec 2008 - 2:52pm
Jack L. Moffett
2005

I'm in agreement with Will, here. The sparklines on this site don't
provide enough context to be of any use. They raise a lot of questions!

1. What is the time-span we are looking at?

2. As Will pointed out, they haven't broken out losses, injuries, and
deaths, which are very different things. If you are a pet owner, a
death is final. An injury is likely mendable, and a loss may only be
temporary.

3. Obviously, a peak represents 1 incident, but each peak covers a
certain span of time. If you look at ATA, which shows 2 incidents in a
single peak with a flat top, and Atlantic Southeast, which shows 1
incident with a pointed peak, the time-span only differs by 2 pixels.
Granted, I don't know the time scale, but it seems to me the width of
a peak is inaccurate in comparison.

4. What does height of a peak represent? Judging by the airlines with
only a few incidents, it has nothing to do with the amount within a
single time unit, as they have full-height peaks. Midwest reports 4
incidents, but there are only three peaks, two of which are half-
height, and one of which is full height. Then look at Continental with
52 incidents. Its sparkline varies in height quite a bit. This leads
me to believe that height is supposed to correlate to the number of
incidents, but the scale is being changed for each sparkline to make
them fit the given height. If such is the case, no meaningful
comparison can be made between them. They may as well just give the
totals and leave it at that.

Will Evans said:
> Agreed, context and scale are important, and breaking out lost,
> injured,
> killed as three separate information vectors by airline, overtime
> without
> connecting the information points would have increased information
> density
> while reducing chart junk.

Actually, Will, chart junk refers to graphics that aren't
communicating data. A sparkline, by definition, is all data. There is
no chart junk. So, breaking out each into a separate vector will
increase the information, but not reduce chart junk. Whether or not it
increases information density depends on how much space is used, of
course. If they are overlaid in the same space, then yes, it would
increase data density.

Best,
Jack

Jack L. Moffett
Interaction Designer
inmedius
412.459.0310 x219
http://www.inmedius.com

There is no good design that is not
based on the understanding of people.

- Stefano Marzano
CEO of Philips Design

22 Dec 2008 - 6:04pm
SemanticWill
2007

Given this was posted to the list today, I find it ironical that Jetblue
sent me a link to this site today
*JetPaws*
http://www.jetblue.com/jetpaws/?source=EMJetpawsAnnounce_main&sp_mid=1318246

On Mon, Dec 22, 2008 at 2:52 PM, Jack Moffett <jackmoffett at mac.com> wrote:

> I'm in agreement with Will, here. The sparklines on this site don't provide
> enough context to be of any use. They raise a lot of questions!
>
> 1. What is the time-span we are looking at?
>
> 2. As Will pointed out, they haven't broken out losses, injuries, and
> deaths, which are very different things. If you are a pet owner, a death is
> final. An injury is likely mendable, and a loss may only be temporary.
>
> 3. Obviously, a peak represents 1 incident, but each peak covers a certain
> span of time. If you look at ATA, which shows 2 incidents in a single peak
> with a flat top, and Atlantic Southeast, which shows 1 incident with a
> pointed peak, the time-span only differs by 2 pixels. Granted, I don't know
> the time scale, but it seems to me the width of a peak is inaccurate in
> comparison.
>
> 4. What does height of a peak represent? Judging by the airlines with only
> a few incidents, it has nothing to do with the amount within a single time
> unit, as they have full-height peaks. Midwest reports 4 incidents, but there
> are only three peaks, two of which are half-height, and one of which is full
> height. Then look at Continental with 52 incidents. Its sparkline varies in
> height quite a bit. This leads me to believe that height is supposed to
> correlate to the number of incidents, but the scale is being changed for
> each sparkline to make them fit the given height. If such is the case, no
> meaningful comparison can be made between them. They may as well just give
> the totals and leave it at that.
>
> Will Evans said:
>
>> Agreed, context and scale are important, and breaking out lost, injured,
>> killed as three separate information vectors by airline, overtime without
>> connecting the information points would have increased information density
>> while reducing chart junk.
>>
>
> Actually, Will, chart junk refers to graphics that aren't communicating
> data. A sparkline, by definition, is all data. There is no chart junk. So,
> breaking out each into a separate vector will increase the information, but
> not reduce chart junk. Whether or not it increases information density
> depends on how much space is used, of course. If they are overlaid in the
> same space, then yes, it would increase data density.
>
> Best,
> Jack
>
>
>

23 Dec 2008 - 3:15am
John Gibbard
2008

Whilst I have used them in Excel reports on metrics (Bissantz
Sparklines plugin) I was (coincidentally) struck by this piece of
information design yesterday when browsing a flickr set. The use of
sparklines here also demonstrates issues with scale though - as I
believe Tutfte points out - scale is less important than trends with
this kind of data.

http://www.flickr.com/photos/juhansonin/393271975/sizes/o/in/pool-575613@N25/

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=36598

22 Dec 2008 - 2:20pm
robfay
2007

A sparkline is good in that it can convey a number of pieces of
information that show a larger trend. This kind of data, however,
does not warrant a trend because there's not enough data to analyze.
Tell me that the data covers 4 years and allow me to sort by count.

The reason a sparkline isn't effective here is because I do not see
any trends or patterns. Either the airline is doing a good job or a
bad job. I cannot see that it's gotten worse or better for the
airline.

There needs to be a few more years of data to work with in order to
start seeing trends.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=36598

23 Jan 2009 - 5:25pm
Casey Edgeton
2008

Here is another neat example. This is a coupon site that shows stats on how
successful the coupon was. This is a nice feature since even though a coupon
has a 90% success rate, the unsuccessful attempts at using the coupon could
either be evenly dispersed or all at the end. This can give you a good idea
of how likely it is that it will work - if all of them are at the end, it
probably has expired.
http://www.retailmenot.com/view/amazon.com

On Mon, Dec 22, 2008 at 10:15 AM, Elise Edson <elise.edson at gmail.com> wrote:

> Hi IxDAers... just thought I'd share this neat example of sparklines!
> I was looking for pet travel information across the various airlines,
> and I was surprised/delighted to see this:
>
> http://www.petflight.com/pet-incidents/airlines
>
> How have you integrated sparklines into your designs?
>
> Elise

24 Jan 2009 - 4:23am
Angel Marquez
2008
24 Jan 2009 - 4:32am
Steve Baty
2009

Angel,

I must be experiencing Saturday night blinker-vision, but I don't see the
sparklines in that site. Where are they?

Steve

2009/1/24 Angel Marquez <angel.marquez at gmail.com>

> http://www.breathingearth.net/
>
>

--
Steve 'Doc' Baty | Principal | Meld Consulting | P: +61 417 061 292 | E:
stevebaty at meld.com.au | Twitter: docbaty | LinkedIn:
www.linkedin.com/in/stevebaty

Blog: http://docholdsfourth.blogspot.com
Contributor - UXMatters - www.uxmatters.com
UX Book Club: http://uxbookclub.org/ - Read, discuss, connect.

Syndicate content Get the feed