thanks for the clarification simon.
according to the manual : "for joins the optimizer bases its estimates for the average number of rows to be returned for each scan of the table and so the total density which provides the average number of duplicates for all values in the column provides the best estimate".
I am not really sure what 'joins' in the above statement means, but for a where clause like this :
where a.col1 = 'Y' and
a.col1 = b.col1
and if the value "Y" already has a frequency weight, why should total density be used for the join?
and what does 'each scan of the table' mean? why should there be multiple scans?
maybe I am not able to visualize the situation - so would appreciate your insights.
also a statistics skew is due to ir-reconciliation between range cell density and total density?
in what kind of situation (join etc) are the two compared?
appreciate the insights.