New opportunities for statistics collection in Teradata 14.0 - comment by carrie

 
The following recommendations for multicolumns statistics are made in my blog posting on statistics collection recommendations for 14.10:
 
Collect Multicolumn Statistics:
 • Groups of columns that often appear together with equality predicates. These statistics are used for single-tables estimates.
• Groups of columns used for joins or aggregations, where there is either a dependency or some degree of correlation among them. With no multicolumn statistics collected, the optimizer assumes complete independence among the column values. The more that the combination of actual values are correlated, the greater the value of collecting multicolumn statistics is in this situation.
• Specify a name for such statistics, for ease of recollection, viewing, and/or dropping.
 
In the absence of column correlation information, the optimizer assumes that columns in a multicolumn statistic are completely independent.  The optimizer assumes that the number of unique values of the join columns doesn’t get reduced after applying the single table predicates.  Further, it assumes that the single table predicates remove the rows evenly from every value group of the join column until there is one row per value. 
 
If single table predicate columns and join columns have a high degree of correlation and the join selectivity estimate by itself is not accurate, then you want to collect base table multicolumn stats with single table predicate columns as leading columns following by join columns.
 
Thanks, -Carrie