I don't have the numbers on hand, but subjectively the number of 'how do I do X' github issues that I deal with has gone down a lot since adopting this style of docs.
Visually this looks very nice. I personally love your choice of colours for the website which compliments the retro look of pattern fills in general. What I'd like more information on, if possible, is whether or not such patterns are more effective, for communication, than regular colours. You're targeting D3 (as opposed to, say the printed page? Please confirm?), so I guess the question is, is this a visual candy (and an effective one for sure), or does it, in addition, have some proof of better effectiveness at visual category demarcation.
I'm not related to the project. I shared a cool link I found online.
But I can think of at least two improvements over using regular colors. The first one being better support for people with color blindness when using a large amount of categories. And the second one is the possibility of showing more than one value at each sector. Something like http://i.imgur.com/SdLeAPa.png shows combination of two kind of data. One being the light/dark background, and the other being the circle/dash pattern
Yes! The color blindness point is a good one. It's amazing how 2 different colors can end up being basically the same when someone is color blind. This image (http://i.imgur.com/4IbhU9w.png) has an example.
Imagine a chart/map/whatever using the 2 colors on the left, but looking to you like the ones on the right. Not that useful.
Actually I was giving a quick talk about this stuff at work using this very example and someone in the room was colorblind and he couldn't tell there was variation in the colors in the sentence. So, basically I got immediate vindication for haranguing the designers.
Tufte dismisses these types of patterns as "chartjunk" and "unintentional optical art"[0]. I'm not sure I agree completely - I think patterns can be used as an effective additional data dimension, but you certainly have to be careful not to overdo it.
>Tufte dismisses these types of patterns as "chartjunk" and "unintentional optical art"
What Tufte doesn't understand is that your end goal isn't always information density and great insight.
Sometimes aesthetics can be equally or more important than the data.
Especially if the data are crap, like 90% of charts used in marketing (), where the marketing part is more important that actually showing something with the chart.
I think tufte understands that people make junk charts on purpose. I think he just doesn't agree that it's a good chart if the purpose isn't to actually inform truthfully . If your data is crap and you know it then basically you are misleading people on purpose. At that point it doesn't really matter if you have a lot of chartjunk as well.
Research in 1984 (Graphical Perception: Theory, Experimentation, and Application to the Development of Graphical Methods - by William S. Cleveland and Robert McGill - https://www.cs.ubc.ca/~tmm/courses/cpsc533c-04-spr/readings/...) found that texture (what they call shading) is worse than color saturation for discrimination of quantitative differences. For categorical discrimination it is probably as good as color if limited to a few patterns.
I see a lot of relatively low level libraries for data visualization. Are there any very high level libraries, that allow regular coders like me to visualize some data quickly? I find myself needing this all the time, but I would prefer the simplest interface possible, rather than high configurability.
If you're familiar with Python, try the Jupyter notebook (jupyter.org) - very useful for quick visualizations while also allowing you to quickly transform the data when necessary. The big advantage is your interactive 'repl' session is persisted as well as editable so you can resume anytime.
Look into http://www.highcharts.com. Much more high level than D3, does nice charts and graphs. Free for non-commercial, and reasonably cheap for commercial work.
The other really useful tool is Tableau Public, if your data isn't confidential. GUI interface, but makes visualizing information really easy.
The other tool I use a lot is http://p5js.org. It's more a general-purpose drawing tool, but if all you want to do is put some boxes at (x,y) positions, it's a really good prototyping tool.
My advice is to bite the bullet and learn d3 (instead of a wrapper like dc.js or nvd3) if this is for any production level data. Or build in the time you will need to rewrite it for API nastiness and probably performance reasons down the line.
I actually modeled the new documentation of Trianglify (http://qrohlf.com/trianglify/) on the Textures.js docs.
I don't have the numbers on hand, but subjectively the number of 'how do I do X' github issues that I deal with has gone down a lot since adopting this style of docs.