tiny JS discretization library for choropleth representation

Question

I'd like to know if there is somewhere a tiny stat/discretization js library ready to use?

Actually I know the existence of jStat (which I may explore a little bit more), and I know mapfish has a geostats module/widget to create choropleth map - I found code here. But it relies on openlayers and I'd like to have a standalone lib and eventually be free to comunicate with openlayers.

Actually, I'd like to have few basics methods :

• min
• max
• mean
• median

and be able to implement the following discretization with the ability to specify the number of classes and the precision (decimal) :

• Quantile
• Equal Interval
• Natural Breaks (Jenks)

If it does not exist yet, I will probably code something by myself and release it on github under free licence. Any resource and advice is most welcome.

Answer 1

Well, I coded something yesterday and released it under MIT License.

The library, named geostats is available on github. The package includes examples. You will also be able to see it in action on http://www.empreinte-urbaine.eu/mapping/geostats/ (with a concrete choropleth representation sample).

It supports the 3 methods listed above :

• Quantile
• Equal Interval
• Natural Breaks (Jenks) : The Jenks classification code has been taken from Doug Curl.
• Unique values

The library also includes a method to render a legend in HTML format.

Quantile classification sample :

Unique Values classification sample :

See a demo reading values directly from shapefile (using Marc Harter library).

If you want to make a try by yourself in 2 steps, using the jsconsole :

Do copy/paste into the console, first :

load("https://raw.github.com/simogeo/geostats/master/lib/geostats.js");

Then, do copy/paste :

var data = Array(12, 22, 5, 8, 43, 2, 34, 12, 34, 36, 5, 21, 23, 45, 12, 31, 38); 
serie = new geostats(data); 
serie.getEqInterval(4);
alert('Classification method : ' + serie.method + "\n\nClasses : " + serie.ranges.join(' | '));

Please, let me know if you discover any bug. Once again, any idea or suggestion is welcome to enhance the library.

That's all! Hope it will also be useful to you.

Answer 2

science.js is a well-implemented collection of stats & linear algebra functionality with no dependencies.

Answer 3

Zachary Forest Johnson has done a really great job to provide thematic analysis on openlayers.

The library is under BSD license, available on github. More explanations on the project and samples are available on his blog.