A Chiasm plugin should be created that wraps the REST API of the data reduction service. In general, this is how visualizations will interact with "Big Data" data sets. For small data sets, the Crossfilter plugin can be used for the same purpose, dealing with data in browser memory.
Deprecated idea. The current approach is to operate on a random sample from a server-side component, then perform the data reduction in the browser via https://github.com/curran/data-reduction
A Chiasm plugin should be created that wraps the REST API of the data reduction service. In general, this is how visualizations will interact with "Big Data" data sets. For small data sets, the Crossfilter plugin can be used for the same purpose, dealing with data in browser memory.