Specifies additional information about the relationship between columns when segmenting a table.




Segmentation by one column may imply segmentation by another column. --segby-advise specifies how those two columns are related. For example, if you have grocery store data and segment by the state column, it implies that it is also segmented by store_id (because the store can exist in only one state). You would then include state and store_id in --segby-advise.

You can use --segby-advise only in conjunction with either --segby or --sortseg. Only one type of segmentation (--segby or --sortseg) can be specified at one time. However, --segby-advise can be repeated and can be used in conjunction with --sortseg-advise.

Note that you can use --segby-advise with --sortseg, to handle cases where a column is segmented the same as the sortseg column, but is not necessarily sorted the same. For example, the state column is sortseg (sorted) but the column store_id is segby (not sorted). In this case, you would use --segby-advise.

For more information about segmentation options, see Segmentation options.