Blog: Dan E. Linstedt« On Hiatus | Main | Data Modeling may also be stuck in 1985! » Stuck in 1985!I've had a bit of time to think while I've been away. I've pondered this question quite a bit as of late. I'm looking into business intelligence presentation layers and wondering why the vendors seem to be stuck in 1985. I'm not talking about the physical data, or the drill through, or the web-capabilities.... No, I'm talking about the graphs and graphics available by the vendors themselves. I find myself asking the question: In this day and age, is it really necessary to "get excited" over executive dashboards that contain the latest bar chart, pie chart, line graph, speed dial, and red-yellow-green coloring? Let me step back for a moment and discuss a few other capabilities that I'm referring to. Take for instance web sites, and web-presence. Most corporate web-sites are moving - full of animation, video, sound, dynamic content and vivid coloring. These days, the old point and click simplistic 2d looking site (mine included) can only hold ones attention for roughly 2 minutes (if I'm lucky). This is one example of the evolution of human-machine interface. Why then must I settle (as an executive officer or senior management) for 2D sterile 3 color bar-charts? Is there something in the corporate culture that prohibits me from having a dynamic 3D fly-through interspatial data exploration experience? Well, there might be... Ok, let's talk about another example. If we move on to Oil and Gas exploration - what kind of interface do the engineers use to explore the ground, for possible oil and gas deposits? How about managing or mapping an existing oil well? For example, check out some of the representations at the bottom of this page (Petroleum Graphics). Granted, geography exists in 3D space - so physically we can represent the data this way according to lat-long, and height or depth. Here's an example of the graphs that they might use for land mapping (RockWorks). There's one particular graph that shows leakage of Radium from tanks, what if we could say the tanks are business units, and the leakage was loss of revenue? Would we be inclined to "pinpoint" the hole in the tank? Would we be interested to know if the hole is the same in all tanks, the same size, location, how many holes? However, why shouldn't we ask the BI vendors to begin producing more sophisticated graphing choices? What if we could get a 3D graph of our business in real-time, where some of the points along the height axis are moving up and down, the 2d access bisected with planes, and colors representing more stagnant components of business? In other words - a graph we can "fly-through" or truly drill into. This type of graph might make BI more fun, dare I say addicting? Imagine how much time we might spend investigating productivity gains or playing with business profitability if motion and fly-through were a part of it... Ok, so there's a lot of mathematics behind the engines, and it may require a deep understanding of the business to assist in developing a custom solution - after all, it's one of the ways a business maintains its competitive edge right? Just think of the visual correlations that could be made with this type of graph, immediate conclusions that were otherwise hidden in two separate 2D graphs. Here's a colored plot (Aspire Software) that begins to be interesting. It represents sound waves and frequencies. Here's another technical plotting (Techplot) software that can represent all kinds of neat real-world elements. Give the page some time to load, then scroll through it. At this point I wonder, is it the business acumen that have been lulled into believing they can only get 2D charts and graph representations from their vendors, or is it the vendors touting that this is the latest and greatest available, and look at this neat coloring feature of this graph? In order to be fair, let me say again - that these vendors have a tremendous amount of fantastic engineering to getting, manipulating, drilling, and walking through the data - it's just the graphing component of the presentation layer that I'm referring to. In all fairness we have to ask: is the average business user ready to make use of all this power on their desktop? Here's an quote from a vendor: Who Uses This Stuff? that you might find interesting: EVS- Environmental Visualization System provides state-of-the art analysis and visualization tools for geologists, geochemists, environmental or mining engineers, oceanographers, archaeologists and modelers. EVS provides true 3D volumetric modeling, analysis and visualization. Advanced site visualization can help reduce site assessment costs and enhance data presentation capabilities for remediation planning, litigation support, regulatory reporting, and public relations. Personally I think its high time big business intelligence and data reporting vendors step to the table, and begin to demonstrate the true power of data visualization. I think the current state of reporting engines is still "stuck in 1985" for the time being. I'm open to comments, thoughts, questions... Again, the BI vendors are strong in the back-end engineering and data integration/retrieval - I'm targeting the graphic presentation specifically. |
Comments
Dan,
You are RIGHT on. I was building an application three years ago, working with a major consulting firm, and went on a "widget search", looking for the best graphics for a dashboard display - 3-D topos, heat maps, drilldowns and bouncing histograms - and what we found was pretty weak. We wound up going with some generic speedometer dials. Doesn't seem like much has changed in the last couple of years....
The best analytical displays I have seen were from an analytics company called Praja, but TIBCO bought them in 2002 and they just disappeared - or at least TIBCO has not been capitalizing on the incredible advantage that would give them. Fortunately, by the time the major BI vendors catch on to this, I think we'll have good graphics available from open source for free. But it's a desert out there right now.
- John Parker
CDIstation.com
Posted by: John Parker | September 7, 2005 6:09 PM
I agree whole-heartedly, too. I think a big part of this is that while we find it straight forward to build a 3D model for something that is 3D in the physical world, we have a tough time representing any higher dimensions... and typical business questions involve many more than 3 parameters. The main data warehouse I work on right now has over a dozen primary dimensions including several degenerate code tables and a few hierarchies. To accurately identify a trend in the data you've got to be able to adjust for known variations in some dimensions and explore several others in parallel. Show me how you do that with a pie chart!? I think the only way to try to visualize that much information is to work with those 3 directional dimensions, then start adding color and animation and icons... etc. Too much information and you overwhelm the user though. What a complex balance to maintain.
As for amazing multi-dimensional imaging technologies, my favorites are the ones that do imaging of real events (rather than models):
GE 4D Ultrasound: http://www.gehealthcare.com/usen/ultrasound/4d/virtual_4d_mini.html
Color Echocardiogram: http://www.cca-austin.com/Services/Echo.htm
Pictures don't even need to be flashy to convey critically important information.
Posted by: Paul Boal | September 7, 2005 10:56 PM