A segmentation rule is used when processing big data to break it into peices (segments), whuch can each be processed largely independently. The rule may be systematic using the natural order in the datastore, time of data, or class segmention when there is some form of cluster or class labels. However, it is also possible to use random segmentation as this may improve the statsitcial properties of later analysis.
Used on Chap. 8: page 162