Question

A) suppose a table. that i want to perform a DELETE function on .
This is done in ms access 2003 sql query. NOTE There are many many many many entries. in the few hundred thousand … to million ranges.so hopefully if there can be a code that deals well with a large dataset. there is 3 types of mood only.

 DayNumber   Mood
     1       Mad
     2       Sad
     2       Happy 
     2       Sad
     3       Sad 
     3       Happy

when there are a few moods in one day we only want to keep the most important one.
so lets have a delete function delete for duplicates of days . first deleting the less important moods. importance of moods is Happy>Mad>Sad. So I want:

 DayNumber   Mood
     1       Mad
     2       Happy
     3       Happy 

B) I was first starting at easier without three options for mood jsut two . where Happy>Sad

 DayNumber   Mood
     1       Sad
     2       Sad
     2       Happy 
     3       Sad 
     3       Happy

Where I will Ideally get

 DayNumber   Mood
     1       Sad
     2       Happy 
     3       Happy

It doesnt matter whether you do the first example or secodn for me I’m stuck either way !

This is what i have for the second question so far.. btu it doesnt work cuz i have an aggregate function in the where clause .

DELETE FROM Table
WHERE (Mood='Sad') and (COUNT(DayNumber)=2);

Answer 1

If you have a small & fixed number of moods, you can hardwire the hierarchy like so:

DELETE FROM Table a
WHERE
(a.Mood='Sad' 
AND EXISTS
  (SELECT 1
   FROM Table b
   WHERE b.DayNumber = a.DayNumber
   AND b.Mood in ('Happy','Mad')))
OR
(a.Mood = 'Mad'
AND EXISTS
  (SELECT 1
   FROM Table c
   WHERE c.DayNumber = a.DayNumber
   AND c.Mood = 'Happy')))