Skip to Content

The traditional means of displaying data using graphs is becoming more and more incongruous to the possibilities offered by new web 2.0 concepts and services.  The ability to bring data to life and free it from static representations, is an increasing trend exemplified by the popularity of mashups and visualization services. 

Decline of graphs

(picture courtesy xkcd)

Everyone is aware of how useful graphs can be in visualizing large amounts of data. Excel is perhaps it easiest and most popular application within enterprises to create charts and visual representations of data. However, the use of these static charting tools to display data reaches a tipping point. Once this stage is reached the law of diminishing returns sets in and looking at more and more graphs begins to lose its appeal and effectivness. After, the third or fourth pie chart in a dashboard or executive summary, people start to lose interest. Also, it’s often the case that charts contain too much information and the associated legends are not particular clear as to what is being represented. In SAP BI this is often the case when the graphical feature is used to display large amounts data within a report. 

The popularity of APIs such as Google-Maps (the top API for mashups) are due to their ability to represent data in new and exciting ways. Representing data layered on a map provides extra context and frees it from traditional representation mediums.  The contrast between overlapping data on a map and merely representing this in a barchart is immense, and can lead to more meaningful analysis and understanding. 

 

The story of Wal-Mart stores

The visualization below tells the story of the growth and expansion of Wal-Mart across America over the past 50 years. The focal point of the presentation is the graphical representation of the data across time. While this data could have been plotted in a bar-chart/line-chart -with time along the x-axis and number of stores along the y-axis – this medium would not tell such a compelling story. The data suits a visual medium, and the map enhances story of the early days of the company i.e. with its origins in Bentonville, Arkansas. The focus on local expansion in its early history is clearly represented, along with its spread in the southeast and then towards the west coast. The sheer speed of expansion is clearly illustrated, and is something that could be glossed over were it represented in a static chart.

For me, the beauty and power of the visualization is outlined by John W Tukey below:

“The greatest value of a picture is when it forces us to notice what we never expected to see” John W Tukey. Exploratory Data Analysis. 1977 (taken from Flowingdata).

 

(visualization courtesy Flowingdata)

The representation above relies on two important criteria – 

1) That the raw data is freely available in an accessible format – This was provided in .csv format at http://www.econ.umn.edu/~holmes/data/WalMart/index.html

2) That this raw data can be overlapped onto a map, with the geographical location of each store plotted based on time. This was provided using Modestmaps

While there is nothing particularly ground breaking about such a mashup, the trend outlined by Google-Maps, OpenStreetMap and others poses a challenge to the presentation of data within enterprises, especially where there is a geographic component. 

Tools such as Visual Composer can already Visual Composer demo – Customer and Google APIAPIs such as that from Google Maps to plot data. However, such services also require a time dimension to creative a cohesive and structured narrative. The Wal-Mart visualization is dependent on time to tell the story of the growth of the chain. The flash technology available within Visual Composer allows for graphs to ‘dance’ when they are manipulated, creating an aesthetically pleasing effect . More of this kind of functionality – including a timeline dimension – is needed in order to create these kind of visualizations within the SAP ecosphere. 

 

Visualizations utilising SAP data

 

As SAP and Business Intelligence systems collect more and more data, exciting and visually appealing ways of presenting this is required. The visual manifestation of this data needs to create a story and tell a narrative in order to be effective. Visualizations are judged on their data, aesthetics and how well they inform the viewer. If the ability to create interesting mashups are left to tools outside the enterprise e.g. Datadepot, Manyeyes, Swivel etc, then there will be more requests for data downloads in csv and xml format from SAP systems. Business users will then take this data and use it with 3rd party services to create the visualizations they require. While this may be a cost effective way to produce interesting presentations, it does introduce security concerns if corporate data is sent outside the firewall to 3rd party applications. Therefore, SAP needs to provide easly means to:

a) allow data to be easily downloaded from SAP systems – including BI in open formats e.g. csv and xml, so it can be used in other 3rd party applications (already possible in BI, ECC and Xcelsius but sometimes extraction is not easy for large amounts of data)

b) enable interesting visualizations of data using third-party maps etc. within SAP e.g. using BI data to create Xcelsius presentations using different APIs and potentially public/benchmark datasets.  

The SAP ecosphere needs to provide businesses with the abilit to create the kinds of visualisations embedded below. Also, it needs to learn from the IBM tools available at Manyeyes which allow for powerful presentations to be created using a simple and easy interface. 

 

Top visualizations

 

The videos embedded below were the best visualizations of the year at Flowingdata. The videos below were taken from the BBC series Britain from Above

(video taken from Youtube)

(video taken from Youtube)

 

Visualization resources 

* Wordle – Creates beautiful word clouds

* Datadepot – Used to upload, share and analyse data and trends

* Manyeyes – Create fantastic and elaborate visualizations

* Swivel –  Create, share and mashup data sets

* Maptube –  Tool for viewing, sharing, mixing and mashing maps online.

 

Visualization examples

 

* Industry Sector Campaign Contributions – uses Google Spreadsheet to highlight contributions over time

* Capitol Words – creates word clouds based on word frequency 

* 5 Best Data Visualization Projects of the year – Visualizations of the year

* OpenStreetMap edits – Visualization of edits on OpenStreetMap

* Google-Maps – Top 100+ Best Tools and Mashups utilising Google-Maps

* StatPlanet –  Demographic, educational, health and other data from various agencies mapped

To report this post you need to login first.

12 Comments

You must be Logged on to comment or reply to a post.

  1. Michael Nicholls
    But it’s only one way to show things. The WalMart pretty picture might look nice, but it doesn’t help tell me how many stores there were in 1980, for example, or give an indication of growth rate, which I can see with a bar chart.
    (0) 
    1. richard fahey Post author
      I agree. If you are looking for the numbers such a visualisation would not be appropriate or useful. I guess my point was that visual interactive presentations are more effective at telling a story than numbers or static charts. The data in the presentation is available within Thomas Holmes’ report at http://www.econ.umn.edu/~holmes/papers/diffusion_walmart_jan2008.pdf (see appendix), but I feel the visualisation is more informative and appealing.

      For many people interested in operational numbers perhaps charts with exact figures are more appropriate. Such visualisations are more suited to higher level presentions where people want a ‘birds eye’ view of things. 

      (0) 
  2. Anton Wenzelhuemer
    …but due to it’s unfortunate design (width of main image I think) combined with the also unfortunate fixed width general layout of this site it is unreadable to me (the text stretches to invisible areas at the right but the scrollbar to scroll there is a mile sown the page. since the text has no margin to the left and right borders it is not possible to center the text once to the available area and then read it but one hast to continuously scroll down to adjust the text, scroll up to read the next line, scroll down again … . unbelievable, we’re in 2009 …).

    I hope this will be improved some time,
    anton

    (0) 
    1. richard fahey Post author
      Hi Anton,
      I am sorry you are having issues viewing the blog. I wrote this on my Mac and checked it in Firefox, Safari and Opera. It looked fine and also the Youtube videos were embedded correctly.

      However, today when I view it on my PC the Youtube videos are not displayed. I agree there is a slight issue with having to scroll over for some of the text. I’m unsure how to fix this though. Can I ask what browser and OS you’re using as I think it reders differently depending on these criteria.

      (0) 
      1. Anton Wenzelhuemer
        hi,

        for your help I tested it on

        • 12″ screen, 1280x 800,Win XP, Firefox 3.0.5
        • 12″ screen, 1280x 800,Win XP, IE 6.0 SP3
        • 18″ screen, 1920x 1080,Win Vista, Firefox 3.0.5
        • 18″ screen, 1920x 1080,Win Vista, IE 7.0

        your böog renders consistently bad on each of them. afaik this is due to the fact that the width of your main embedded object (800 px) is too large for the general layout of this site.

        the whole textual content then adjusts to the (too large) width of the SWF-‘image’.

        my 2 cents,
        anton

        (0) 
        1. richard fahey Post author
          Thanks for that information. I have reduced the layout of the .swf file to 600×300. Hopefully this works better and apologies for the inconvenience.
          (0) 
  3. Anonymous
    I am huge fan from social networking, so as a logical consequence I became a fan of TouchGraph (http://www.touchgraph.com/navigator.html) For those that don’t know TouchGraph yet, Touchgraph is a powerful visual mapping tool that allows you to discover the relationships contained in information sources, including cluster detection and centrality ranking. Muckety.com uses simelar techniques: http://news.muckety.com/

    Being responsible for market development in the public area, I am used to build influence maps (using powerpoint). My influence maps vizualize reationships between my customer, the market and our employees and help me to define, track and present my influencing activities more efficiently.

    It would be great if we could automaticly generate such visualisation based on information stored in SAP’s CRM. This would not only make sales activities more efficient, but encourage me to even enter more data in our CRM system.

    what do you think?

    (0) 
  4. Larry Griffiths
    Although there are numerous  and exciting new ways of displaying data what I find most frustrating is the continued poor design of Visualizations by BI developers/analyst. All the bells and whistles in the world can’t make up for poor design.  

    That said, if you work in the BI space and haven’t read “Show me the Numbers” by Stephen Few go out and do so today. You can find his web site below with a number of interesting links, blog entries, and articles. Good reference for anyone one who has to deal with having to display data.
    http://www.perceptualedge.com/

    (0) 

Leave a Reply