On Aug 27, 2006, at 1:59 PM, Lars Jensen wrote:
Can you share some examples of attractive graphs?
wow, that's a challenge. It's too tough to do in a short message, so instead I'll give some thoughts and personal practices.
First, from my point of view, all graphs should use vector graphics exclusively. These are almost essential for creating publication- quality graphs.
Second, when I make graphs I make them with the idea of exporting, through the clipboard, to a graphics layout program such as Illustrator (which I don't use) or Canvas (which I do use). I do this whether I am using a devoted graphing program (such as kaleidagraph) or a graph made from REALbasic. Fortunately, REALbasic does an adequate job of putting vector graphics into the clipboard. I don't think it realistic to ask REALbasic to give the flexibility of a graphics layout program for nudging groups of vectors, adding text, changing colors, alignment, conversion of vectors to bitmaps, etc. I think it is asking far to much of REALbasic to create final-quality graphics.
Let me give an example. Our group recently submitted a paper to the journal of Neuroscience. Most of the figures were composite figures. Figure 8 contained 8 graphs. Each was transfered, via the clipboard, to Canvas. Since these were all vectors graphics pictures, I had a lot of flexibility (which I used). There were lots of dots of different colors. We decided to change the size and color of some of the dots. In Canvas, I could search and select all of the dots of a particular color and then change their fill and and size. In the composite canvas document I could align the graphs, add text, re- arrange and, as a final step, convert to bitmaps (which the journals like).
So, what I would ask of a scientific program with strength in graphing is the ability to create component graphs.
Third, I believe attractive graphs are simple and uncluttered. One of the trickiest graph types, and probably 1/3rd of my graphs, are histograms (not to be confused with bar graphs). The tricky thing about a histogram is getting the bin size right, getting the graphics limits right, and deciding whether the bins should be central or left edge. Scaling and labeling the x-axis is also tricky.
My taste is to avoid strange color fills or strange symbols. Colors are for differentiation, not decoration. (My histograms are always black and white)
The great advantage of REALbasic over a graphics package is total flexibility. I would guess that most scientific and engineering presentations are a mix of standard graphics types and home-grown types. The artistry of graphics presentation is making creative and informative home-grown plots. In my work I keep track of events that occur in specific locations (2d). I've used REALbasic to make a variety of plots of this type of data. None are standard plots. As a rule, I use graphing programs for standard graphs and REALbasic for the homegrown stuff. But the extra steps involved in going thru a graphing program can make this awkward, and I have tried to make a general purpose histogram grapher.
The challenge of giving examples of "good" and "attractive" graphs is too difficult. To do an adequate job would probably take about a dozen examples and a fair amount of explanation. A beginning is to look at top-flight scientific journals, such as Science or, in my field, Journal of Neuroscience. But even in these there are many poor plots. The criteria I look for: simplicity and lack of clutter; attractive layout; easy to read standard plot; interesting and attractive home-grown plots.
John Kubie _______________________________________________ Unsubscribe or switch delivery mode: <http://www.realsoftware.com/support/listmanager/> Search the archives of this list here: <http://support.realsoftware.com/listarchives/lists.html>
