I wrote the Clustered-Stacked Column Charts post based on Excel 2003, but in my comment of September 5, 2008, I described changes to the protocol in 2007.

Agree with Chrisham – making your training materials available on-line for a fee to those that don’t have the ability to attend your courses might be a good addition to your business model. You might also consider offering your entire suite of utilities as one piece of software – similar to JWalk’s Pup7. Finally, have you ever considered writing a book? To start you could self-publish an ebook and make it available on your site for a fee. –Just some thoughts. Have a good New Year.

John

Happy New Year

On the text elements side, what about a utility that truncated text elements that word-wrapped, so that they didn’t word-wrap?

“If they are too close, it turns black.” True. Or even worse, if they are solid blocks (e.g. squares with black fill) then you might get a whole lot lumped together and you would never know how many are underneath…which is why I used jittering (i.e. shapes with an outline but no fill, so you can better see overlaps ) in the example.

But I think they do give a good enough ’sense’ of density for the purpose of working out where are the thin parts and where are the thick parts of the distribution. Far from precise, but still a handy way to look at the underlying data in a histogram that might otherwise hide some hotspots. A visual analogy would be looking down on say a fairground crowd: Where is the highest density of people? At the front of the best attraction; in the lines to use the toilets; and at the bar. How many discrete people are there at each point? Don’t know, but it looks like there’s fairly equal density measures between the bar and the toilet for some strange reason ;-)

THat said, a cumulative probability chart is an excellent idea. Only downside is that it might take more space on the chart, and non chart-geeks might not understand it as well as a 1-d frequency chart.

(Random aside: All this talk of dimensions made me think of the movie Avatar which I saw the other day…I agree with the critic who said it’s such a pity a 3-D visual masterpiece has characters who are soooo completely 1-D)

Replacement text elements: Microsoft has promised that they would introduce user-sizable text elements, but then 2007 didn’t have them, nor does 2010. I wrote a routine once to replace category labels on a bar chart with text boxes because the labels were too long and wrapped. The logistics are tricky, to get a text box or other shape to stay aligned with a particular XY pair without actually locking it to a point. But it’s worth thinking about.

One-dimensional frequency chart: I don’t like these, because you really don’t get a good sense of the density of points. If they are too close, it turns black. If they are too sparse, they’re hard to read. They are essentially using shades of gray to encode values, or worse, they force me to get out my ruler and count how many points fall within each cm along the axis. What’s wrong with a cumulative probability chart? Even if your Y axis is counts instead of Z-scores, it’s easy to read. Steep=Dense.

Custom number formatting for chart axes: Well, I could get into that one. There could be some “standard” ones, then some user-saved ones. A little dialog could help design a format. I’ll bet Naomi would like that one, because it would cut down on repetitive %, $, etc.

How about an axis title manager, which you store commonly used axis titles, and you can apply any title from a dropdown.