Click here to monitor SSC
SQLServerCentral is supported by Red Gate Software Ltd.
 
Log in  ::  Register  ::  Not logged in
 
 
 
        
Home       Members    Calendar    Who's On


Add to briefcase 12»»

Visualizations Expand / Collapse
Author
Message
Posted Thursday, May 9, 2013 10:15 PM


SSC-Dedicated

SSC-DedicatedSSC-DedicatedSSC-DedicatedSSC-DedicatedSSC-DedicatedSSC-DedicatedSSC-DedicatedSSC-DedicatedSSC-DedicatedSSC-DedicatedSSC-DedicatedSSC-DedicatedSSC-Dedicated

Group: Administrators
Last Login: Yesterday @ 6:53 PM
Points: 31,177, Visits: 15,623
Comments posted to this topic are about the item Visualizations






Follow me on Twitter: @way0utwest

Forum Etiquette: How to post data/code on a forum to get the best help
Post #1451406
Posted Friday, May 10, 2013 1:55 AM
SSCrazy

SSCrazySSCrazySSCrazySSCrazySSCrazySSCrazySSCrazySSCrazy

Group: General Forum Members
Last Login: 2 days ago @ 7:36 AM
Points: 2,907, Visits: 1,832
I've just been on a data visualisation workshop and we were shown visuals that really blew my mind.

Then the presenter said, "now tell me what the visual is actually telling you, extract the useful facts".
His point was well made. A lot of visuals are very pretty, often entertaining but in giving precise actionable facts they lag behind.

He talked about something called the Data to Ink Ratio.

  • = data ink / total ink used to print the graphic

  • = proportion of a graphic’s ink devoted to the non-redundant display of data-information

  • = 1.0 − proportion of a graphic that can be erased without loss of data-information



The amount of information a human can take in is limited so 3D charts with multiple colours might look pretty but they actually distract from the true information content.

He listed a number of principles
1. Display neither more nor less than what is relevant to your message
2. Do not include visual differences in a graph that do not correspond to actual differences in the data
3. Use the lengths or 2-D locations of objects to encode quantitative values in graphs unless they have already been used for other variables
4. Differences in the visual properties that represent values (that is, differences in their lengths or 2-D locations) should accurately correspond to the actual differences in the values they represent.
5. Do not visually connect values that are discrete, thereby suggesting a relationship that does not exist in the data
6. Make the information that is most important to your message more visually salient in a graph than information that is less important
7. Augment people’s short-term memory by combining multiple facts into a single visual pattern that can be stored as a chunk of memory and by presenting all the information they need to compare within eye span

During the workshop we were given various tests that demonstrated the validity of the principles. One of the exercises was to look at a typical Excel chart and subtract all the irrelevant parts from it.
On a 3D bar chart the stuff that was removed was

  • 3D affects, 2D communicates facts better

  • Borders around legends, bars and the graph itself

  • Tick marks on vertical axis. The human brain is very good at working out differences in length and height and if you want precision you can always go back to the raw data

  • Data labels above bars. Again the human brain gives a very good approximation

  • Legend key in combination with data labels. You've got the legen in any case[li]
    [li]Vertical background lines. They serve no purpose on a bar chart

  • Grey background

  • Bold/Underline text on items not requiring emphasis. Don't use bold on axis labels.



Apart from extending the life of your inkjet cartridge the facts in the resulting chart simply jumped out at me. Less really was more!

There is real science behind the points raised. One example given was the image used prior to the Challenger disaster that was used to recommend that the shuttle not be flown in excessively cold conditions. The relevant information is in the graph if you spend the time to look at it. If a simple plot of faults vs temperature had been plotted the risk of launch in low temperatures would have jumped out.



LinkedIn Profile
Newbie on www.simple-talk.com
Post #1451461
Posted Friday, May 10, 2013 3:02 AM


SSC-Enthusiastic

SSC-EnthusiasticSSC-EnthusiasticSSC-EnthusiasticSSC-EnthusiasticSSC-EnthusiasticSSC-EnthusiasticSSC-EnthusiasticSSC-Enthusiastic

Group: General Forum Members
Last Login: Friday, October 17, 2014 1:16 AM
Points: 106, Visits: 782
David.Poole (5/10/2013Then the presenter said, "now tell me what the visual is actually telling you, extract the useful facts".


First point is we seem to quickly forget that numbers and text are actually an extremely accurate and well defined data visualization methodology which have been developed over centuries (getting on for millenia). Unusually because of our exceptional shared understanding of their intricacies and familiarity few of us even consider it visualization anymore.

Secondly I wouldn't underestimate the importance of colour coding things like straight tables to improve understanding. Things like green numbers to highlight rises and reds for drops can brilliantly allow us to focus in on numbers. In situations where quick decisions need to be made or you are constantly monitoring streams of information it can be invaluable for focusing attention.

Given the above points I would first say that most visulizations should be developed slowly over time with plenty of time and effort put into exactly how they look and the way they can be interacted with. Zooming in and Zooming out might be important along with allowing people to trigger access to detailed figures from the graphs. Also I like to see them continued as a reference point and hopefully maintained in some kind of live feed basis. People are not familiar with graphs / colour indications etc the way they are familiar with text and figures so you need to give your users time to get familiar with what is being shown.

I have no doubt that suitably understood non standard graphic representations of data are an extremely worthwhile and useful tool that is a valuable addition for users. I hope to include more in my day to day work. Maybe we just need the same amount of effort spent on familiarzing ourselves with data visualization techniques as we've spent on numbers and text symbology.

I would add that if you are dealing with any land parcel administration system - geographical visualization is not just an add on. The geographic location is actually the primary key and because of its complication in numbers can only be accurately summarised to users by visualization on a map. People will refer to land parcels in multiple ways (plot no, address, field no, postcode) but the only way to truly cross reference such records is to show the accurate boundary on a map and thereafter overlay them against each other.
Post #1451476
Posted Friday, May 10, 2013 6:57 AM


SSC Veteran

SSC VeteranSSC VeteranSSC VeteranSSC VeteranSSC VeteranSSC VeteranSSC VeteranSSC Veteran

Group: General Forum Members
Last Login: Tuesday, July 8, 2014 1:39 PM
Points: 245, Visits: 735
Really good article and follow-up comments. Thanks a lot.

It is always really scary to me when Management look at pie charts and graphs and "Think" they understand the numbers. All they really understand is what the chart is telling them. Hopefully the presentation was well thought out. Too often the scenario is "I need a chart for a meeting in 2 hours" ...


<><
Livin' down on the cube farm. Left, left, then a right.
Post #1451563
Posted Friday, May 10, 2013 7:03 AM
SSC-Enthusiastic

SSC-EnthusiasticSSC-EnthusiasticSSC-EnthusiasticSSC-EnthusiasticSSC-EnthusiasticSSC-EnthusiasticSSC-EnthusiasticSSC-Enthusiastic

Group: General Forum Members
Last Login: Yesterday @ 3:07 PM
Points: 169, Visits: 1,853
This article has a nice list of tools for data visualization:

http://www.netmagazine.com/features/top-20-data-visualisation-tools


and the person I've follow the most is Edward Tufte, http://www.edwardtufte.com/tufte/

i enjoy his moderated forum: http://www.edwardtufte.com/bboard/q-and-a?topic_id=1
Post #1451567
Posted Friday, May 10, 2013 7:40 AM
Forum Newbie

Forum NewbieForum NewbieForum NewbieForum NewbieForum NewbieForum NewbieForum NewbieForum Newbie

Group: General Forum Members
Last Login: Friday, May 9, 2014 3:06 PM
Points: 2, Visits: 27
Both Edward Tufte and Stephen Few are excellent for very different reasons.

Tufte is more the theorist and artist. Few the practioner. I've been to both of their classes/trainings. I typically appreciate and lean toward abstraction and theory. However, I strongly urge that if you could only go to one class or buy one book, that it be a class or book by Few.
Post #1451593
Posted Friday, May 10, 2013 8:23 AM


SSC Veteran

SSC VeteranSSC VeteranSSC VeteranSSC VeteranSSC VeteranSSC VeteranSSC VeteranSSC Veteran

Group: General Forum Members
Last Login: Yesterday @ 6:14 AM
Points: 272, Visits: 1,354
I've seen quite a few demos (esp. involving Power View) where the visualization being demonstrated looks fantastic, but it actually doesn't say anything useful (or takes up a lot of space to say very little).

When it comes to reports & visualizations, the most basic question I ask is "What action will you take based on this report?". Often people have very fixed ideas of what they want to see, and this question gets past that and helps me ensure that the final result meets that goal.


Leonard
Madison, WI
Post #1451621
Posted Friday, May 10, 2013 10:06 AM
SSC Veteran

SSC VeteranSSC VeteranSSC VeteranSSC VeteranSSC VeteranSSC VeteranSSC VeteranSSC Veteran

Group: General Forum Members
Last Login: Yesterday @ 12:04 PM
Points: 266, Visits: 2,601
David Poole: Wow! That's great info. Sounds like a fantastic class. Thank you for taking the time to write as much of it as you did and to include the story at the end. That sent chills down my spine.
Post #1451676
Posted Friday, May 10, 2013 10:18 AM
SSC Veteran

SSC VeteranSSC VeteranSSC VeteranSSC VeteranSSC VeteranSSC VeteranSSC VeteranSSC Veteran

Group: General Forum Members
Last Login: Yesterday @ 12:04 PM
Points: 266, Visits: 2,601
I'm not sure this is what the post is talking about, but I use visuals for myself when trying to figure out more complex queries--both existing ones someone else created and new ones that I need to create. I sketch the tables/views on paper with just the relevant information and lines. I (sometimes) find it very helpful.

Also, while not an image which: "condense information into a much smaller space.", I find giant, poster size printed data models (when laid out and color coded in a helpful way) to be very helpful in me being able to grasp or simply use a database.

As for helping business people find tools or methods to work with data: I'm mostly of the mind that the vast majority of lay people are not qualified (on a couple levels) to create their own reports. I think it is in everyone's best interest to have a professional put the reports together that the business people ask for. I think MS Access works especially well for this purpose because you not only get a great set of reports, but you can design an appropriately sophisticated user interface for accessing the reports. The need for a good user interface becomes more important the greater the number of reports that are needed.
Post #1451680
Posted Friday, May 10, 2013 10:38 AM
SSC-Enthusiastic

SSC-EnthusiasticSSC-EnthusiasticSSC-EnthusiasticSSC-EnthusiasticSSC-EnthusiasticSSC-EnthusiasticSSC-EnthusiasticSSC-Enthusiastic

Group: General Forum Members
Last Login: Thursday, September 11, 2014 9:56 AM
Points: 145, Visits: 593
stephen.lloyd 63174 (5/10/2013)
Both Edward Tufte and Stephen Few are excellent for very different reasons.

Tufte is more the theorist and artist. Few the practioner. I've been to both of their classes/trainings. I typically appreciate and lean toward abstraction and theory. However, I strongly urge that if you could only go to one class or buy one book, that it be a class or book by Few.


I agree completely. I've been also to their classes, and to others' as well, and working well with data is a great blend of both art and science. Tufte's books explore that very well.

The whole topic of data visualization is a fascinating one, and lately I've come to think of it as at least as important as the db side of things. The overview of techniques article was a good one as a intro into the many different approaches possible. That same Stanford lab has produced a number of pioneers in the field (Tableau is one result, but there are also good competitors) that have made it very easy to explore data without programming skills. The premise is that much can be revealed from data by selecting the appropriate visualization tool. The catch is that it is very hard work to do it well (like most things, no?), and it requires a ruthless kind of honesty to be clear when one is just making a pretty picture. But it is a very enjoyable way to work with data, bring an aesthetic sense to data work, and end up with something simple and elegant (other than a great data model or sql query, which few can appreciate). By the way, an easily accessible source of highly regarded visualizations is the New York Times -- their work is often singled out for being high quality. Also Hans Rosling's TED talks are famous in this regard.



Post #1451687
« Prev Topic | Next Topic »

Add to briefcase 12»»

Permissions Expand / Collapse