I enjoy the interaction as well. I hope you still like it after you read my comment to your follow up post. The defragged partition chart reminds me of the square pie charts which were all the rage last year. Defragging the partition chart helps somewhat, but I think it’s still less effective than the pie I posted in my article (Ineffective Chart – Partition Chart ). And the pie was less effective than the bar chart, although a three-point pie chart isn’t terribly bad.

I generally try to avoid vba and try to stretch the limits of existing functionality to get the work done. thanks for the feedback, keep visiting

@Jon .. I totally agree, conditional formatting is a powerful tool. I have written about it here – http://chandoo.org/wp/2008/03/13/want-to-be-an-excel-conditional-formatting-rock-star-read-this/ where I also spoke about the gantt chart hack, providing the feedback etc. thanks for the discussion, I enjoy this

I should say, the ability to use conditional formatting (CF) is important. You may not have met your objectives with this chart, because as you say, it was random data used to describe an approach. This approach would be reasonable as you said for click tracking, or for eye tracking studies that show where visitors to a site focus their attention.

Not too different from this approach is using CF to draw a simple Gantt chart. It’s the first technique I show on my tutorial on Gantt Charts. CF has many other uses as well, not just in presenting data, but providing feedback to the user, for example, when they’ve entered an invalid piece of data in a cell.

Thanks again. I love your excel tips. I am also kind of pro in excel but i tend more towards VBA. I guess that’s why i like your simple excel tricks. keep up the good work!!

@Stephen … You have an awesome point, but I didnt bother to do any trend (not time series though, this is mostly spatial) because I used random data (generated a range of random numbers from 1-100 and assigned 3 colors based on the value, that is why it clearly looks like garbage or as Jon said, a bunch of M&M’s)

I think my example of city-wise top selling products is slightly misleading as it requires the data to be arranged based on the real lat / long of cities (which I didnt mention).

Let me give you a better / relevant example. Take the user mouse clicks data for a sample web page (1024*768) and suppose you have the click data for each pixel over some time period (lets say 24 hours) then this kind of chart can clearly reveal where user is clicking most – a la a heat map.

I guess, I have focused more on “How-to” part of the chart to the extent that I neglected the “why / what is” part of it. Its good to have this discussion so that I can learn / clarify these points.

and btw, welcome to PHD Blog, hope you like it here

In your response to Jon Peltier, you said that your example of a partition chart reveals trend information. The presence of meaningful trend information isn’t obvious to me. What are the trends that you discern from this chart and what are the visual patterns that reveal them?

The term “trend” usually refers to the overall direction and amount of change through time, but your partition chart doesn’t appear to show change through time. If each cell represents a different city, as you have suggested, and the color represents which of three products sold best in each city, assuming that the cities are arranged alphabetically, the positions of each of the three colors are meaningless. There are not only no trends to observe, but there are no meaningful patterns of any kind.