Skip to main content
. Author manuscript; available in PMC: 2016 Dec 23.
Published in final edited form as: Stat Sci. 2014 Aug 18;29(2):214–226. doi: 10.1214/14-STS476

Fig. 2.

Fig. 2

Grouping via partitioning vs. splitting into multiple objects. Top: the input vector, with elements belonging to three different groups: red, blue and yellow. Middle: typical splitting of vector into three vectors, one per group. This brings the overhead of multiple objects. Bottom: the data are virtually split by a partitioning, encoded by the number of elements in each group (the vector is assumed to be sorted by group).