Filling out a data set

[deleted_user]

I have some fisheries data that has what species was caught for a particular sampling effort, but not what species were not caught; i.e. not all species were caught during every effort, and only those that were caught were put into the data set. Therefore when density is calculated it only includes those efforts where there was a positive instance, thereby over inflating the density estimate. How would I go about working the data step so that the density calculation per species includes a zero for those times where that particular species was not caught+

[LinusH]

An idea:
First create a seperate table containing every possible combination of the PK columns (i.e. data, location, species): do a select distinct on your transactional data on the key columns (except species). Then create a cartesian product table by joining the result to a species table (containg all your species) without specifying join columns.

Now do a full join/merge with this table. In this step you can specify 0 for those rows populated from the cartesian table alone.

[deleted_user]

You could us "sparse" option in frequency procedure in producing 2 dimensional table. then use freq statement in KDE procedure.