SQL select max(date) and corresponding value [duplicate]

Question

Possible Duplicate:
How to get the record of a table who contains the maximum value?

I've got an aggregate query like the following:

SELECT TrainingID, Max(CompletedDate) as CompletedDate, Max(Notes) as Notes     --This will only return the longest notes entry
FROM HR_EmployeeTrainings ET
WHERE (ET.AvantiRecID IS NULL OR ET.AvantiRecID = @avantiRecID)
GROUP BY AvantiRecID, TrainingID            

Which is working, and returns correct data most of the time, but I noticed a problem. The Notes field which gets returned will not necessarily match the record that the max(completedDate) is from. Instead it will be the one with the longest string? Or the one with the highest ASCII value? What does SQL Server do in the event of a tie between two records? I'm not even sure. What I want to get is the notes field from the max(completedDate) record. How should I got about doing this?

Enstatite

Is CompletedDate not a DateTime? – Radack

It is a DateTime. No problem with that field, but with Notes. – Enstatite

the max notes or every notes for the max(completedDate)? – Daves

I used Max(notes) to try and get what I want, but I don't actually want the max of notes. I really want whichever notes is in the same record as the max(dateCompleted). – Enstatite

Answer 1

You can use a subquery. The subquery will get the Max(CompletedDate). You then take this value and join on your table again to retrieve the note associate with that date:

select ET1.TrainingID,
  ET1.CompletedDate,
  ET1.Notes
from HR_EmployeeTrainings ET1
inner join
(
  select Max(CompletedDate) CompletedDate, TrainingID
  from HR_EmployeeTrainings
  --where AvantiRecID IS NULL OR AvantiRecID = @avantiRecID
  group by TrainingID
) ET2
  on ET1.TrainingID = ET2.TrainingID
  and ET1.CompletedDate = ET2.CompletedDate
where ET1.AvantiRecID IS NULL OR ET1.AvantiRecID = @avantiRecID

Jim

Answer 2

Ah yes, that is how it is intended in SQL. You get the Max of every column seperately. It seems like you want to return values from the row with the max date, so you have to select the row with the max date. I prefer to do this with a subselect, as the queries keep compact easy to read.

SELECT TrainingID, CompletedDate, Notes
FROM HR_EmployeeTrainings ET 
WHERE (ET.AvantiRecID IS NULL OR ET.AvantiRecID = @avantiRecID) 
AND CompletedDate in 
   (Select Max(CompletedDate) from HR_EmployeeTrainings B
    where B.TrainingID = ET.TrainingID)

If you also want to match by AntiRecID you should include that in the subselect as well.

Mag

Answer 3

There's no easy way to do this, but something like this will work:

SELECT ET.TrainingID, 
  ET.CompletedDate, 
  ET.Notes
FROM 
HR_EmployeeTrainings ET
inner join
(
  select TrainingID, Max(CompletedDate) as CompletedDate
  FROM HR_EmployeeTrainings
  WHERE (ET.AvantiRecID IS NULL OR ET.AvantiRecID = @avantiRecID)
  GROUP BY AvantiRecID, TrainingID  
) ET2 
  on ET.TrainingID = ET2.TrainingID
  and ET.CompletedDate = ET2.CompletedDate

Fantasize

you syntax is wrong for your subquery, you are missing a FROM clause. – Jim

Answer 4

Each MAX function is evaluated individually. So MAX(CompletedDate) will return the value of the latest CompletedDate column and MAX(Notes) will return the maximum (i.e. alphabeticaly highest) value.

You need to structure your query differently to get what you want. This question had actually already been asked and answered several times, so I won't repeat it:

How to find the record in a table that contains the maximum value?

Finding the record with maximum value in SQL

Holism

If the question has been answered many times before, the correct thing to do is to vote to close as "exact duplicate" – Straight

Thx, I've now done that. – Holism