Introduction to Tableau and Spotfire
From Visualization Sp06
Lecture on Jan 26, 2006
Readings
- Polaris (extended paper), Stolte, Tang and Hanrahan (pdf)
- Dynamic queries, starfield displays, and the path to Spotfire. Schneiderman (html)
- Chapter 8: Data Density and Small Multiples, In The Visual Display of Quantitative Information. Tufte.
- Chapter 2: Macro/Micro Readings, In Envisioning Information. Tufte.
- Chapter 4: Small Multiples, In Envisioning Information. Tufte.
Contents |
Pushkar - Jan 30, 2006 06:04:51 pm
This is a high-level question that concerns the use of Tableau and Spotfire.
Both programs demonstrate the ability to show a high-level picture of the data, which in most cases isn't very helpful (unless we get the visual parameters just right). Alternatively, both are able to evaluate SQL-like queries (which could also be done by a database tool). To perform lower-level queries effectively, one has to be familiar with the data, which might make its visualization unnecessary. If one is not familiar with the data, it'll take some trial and error before being able to come up with the right visualization parameters (just as a database designer would have to try various SQL queries).
My question: what the ideal user demographic of this type of visualization software? Database designers familiar with the data who wish to show its characteristics? Or data users unfamiliar with the data who wish to understand it?
Bryan - Jan 31, 2006 10:23:33 am
Similar to Pushkar's questions, but perhaps taking things a step further: It seems to me that, like he said, it requires special knowledge of the data model in order to effectively visualize the data. If special knowledge is already available, then tools like these seem unnecessary; you could assemble something by hand using gnuplot that exactly fit your intended audience and the point you're trying to make.
On the other hand, with no knowledge of the data, it seems likely that you could lead yourself astray just playing around with one of these programs, and end up with an unintentionally misleading visualization. Maybe all that is demanded is a slightly better example. I can imagine a researcher in a pharmaceutical company who knew all about her data, but who was unable to draw conclusions about it because there was simply too much of it without visualization. In this case, I can understand the use of these programs. Maybe it would be good in the future to demonstrate something like this (data mining, massive aggregation, non-obvious row-column relationships)?
Yi-Tao - Jan 31, 2006 01:51:42 pm
I agree with Bryan. The examples from lecture didn't present many options; there just wasn't enough specific questions you could ask about the data. Since the majority of data points were concentrated into specific regions, we will obviously only notice the outliers.
As for using his other comment about using GNUplot to generate specific charts, I got the impression that Tableau and Spotfire were more geared to that purpose. They made the process of creating visualizations easy but they didn't prescribe the correct ones; although Spotfire seemed to attempt to do so.
Raymond - Feb 01, 2006 10:02:18 pm
In regard to Brian's comment, I thought that both Tableau and Spotfire was proivde more flexibility while still maintaining an ease of use. I also thought their intended audience would be business/corporate office employees, who are more likely to use Excel to make visualizations instead of GNUplot.
In my opinion, Tableau seemed to be the better product but both have improvements to be made (such as layering, which was brought up in class).
Nchentan - Feb 02, 2006 12:21:39 am
I agree with Raymond's comment in the regard that the intended audience are the busines employees who are familiar with Excel rather than for scientists or researchers, because it gears more toward user friendliness than functionality. As demonstrated in class, where we ask question about data, observe the resulting plot, notice some interesting feature about it, and then change the question and repeat the process again to try to observe the interesting feature more closely. The programs make this loop fast and more user friendly than to change the SQL query, and then create a plot from the resulting view by many commands. I would also argue that they may be the right tool to use, when you want to find out if there are some patterns in the data which involves lots of trial and error.
Mehershad - Feb 02, 2006 12:39:44 am
One of the features I commonly find missing in graphical visualization tools is to create a break and continuation of a bar graph. I did not see the feature in either of the softwares presented today. Putting aside the arguments against doing such a visualization, I have often encountered cases when 1 elements value is so large that the y scale adjusts to its value and hence the other values look irrelevant. In those cases it would be worthwhile to create a crack sort of in the bar so that we can still see the other values. Please correct me if I am wrong because we didnt get an indepth look at Tableau or Spotfire.
Brien - Feb 02, 2006 01:52:13 am
I think you're right, Bryan, that this software really shines with a large data set. NASA provides data from some of its missions at ftp://nssdcftp.gsfc.nasa.gov/ (in textual .asc files, space separated. I had to run them through a regex to replace runs of spaces with a comma before loading it into Spotfire). Spotfire may not be as flexible as Tableau, but it's pretty easy to use. I like how each slider can operate in different filter modes, via a right click. I also agree with you, Nuttapong, that this software beats writing SQL to dig into data.
Sharena - Feb 02, 2006 10:36:57 am
Thinking about the user for these kinds of systems does bring up an interesting point. In the reading on Polaris, it says that it can be used for anything from data mining to scientific computing. Still, most of the examples in the paper are business related. And, it seems to me that any user can create misleading graphs using any system, but I suppose using these tools are so fast if you don't pay attention to what you are doing it is easier to create graphs that do not really mean anything, but that also means that you can see incorrect graphs more quickly if you know something about the data relationships. On another note, that paper on Polaris talks about Polaris including layering, but it seems this may have not been incorporated in Tableau, as of yet perhaps?
