分组其他字段时，SQL CASE和CAST一起导致聚合错误

Question

I have a query that has to convert values into decimals and divide them. I need to only divide when a value is greater than 0, otherwise it throws an error. My query is below. For some reason adding the CASE statement causes an aggregation error. I've seen examples that say to wrap in a SUM, but that won't work for my case. Any suggestions?

SELECT
    EmailName,
    SUM(Sent) as Sent,
    SUM(Delivered) as Delivered,
    SUM(UniqueOpens) as UniqueOpens,
    SUM(TotalOpens) as TotalOpens,
    SUM(UniqueClicks) as UniqueClicks,
    SUM(TotalClicks) as TotalClicks,
    SUM(Unsubscribes) as Unsubscribes,
    CAST(SUM(UniqueOpens) AS FLOAT)/CAST(SUM(Delivered) AS FLOAT) as OpenRate
FROM 
    [Non Triggered Sends Last Month]
GROUP BY 
    EmailName

This works if UniqueOpens is greater than 0, but fails otherwise

This throws an aggregation warning

SELECT
    EmailName,
    SUM(Sent) as Sent,
    SUM(Delivered) as Delivered,
    SUM(UniqueOpens) as UniqueOpens,
    SUM(TotalOpens) as TotalOpens,
    SUM(UniqueClicks) as UniqueClicks,
    SUM(TotalClicks) as TotalClicks,
    SUM(Unsubscribes) as Unsubscribes,
    CASE WHEN UniqueOpens > 0 THEN CAST(SUM(UniqueOpens) AS FLOAT)/CAST(SUM(Delivered) AS FLOAT) ELSE CAST(SUM(UniqueOpens) AS FLOAT) END as OpenRate
FROM 
    [Non Triggered Sends Last Month]
GROUP BY 
    EmailName

An error occurred while checking the query syntax. Errors: Column 'C7211451.Non Triggered Sends Last Month.UniqueOpens' is invalid in the select list because it is not contained in either an aggregate function or the GROUP BY clause.

Answer 1

I think you just need to SUM(UniqueOpens)

SELECT
EmailName,
SUM(Sent) as Sent,
SUM(Delivered) as Delivered,
SUM(UniqueOpens) as UniqueOpens,
SUM(TotalOpens) as TotalOpens,
SUM(UniqueClicks) as UniqueClicks,
SUM(TotalClicks) as TotalClicks,
SUM(Unsubscribes) as Unsubscribes,
CASE WHEN SUM(UniqueOpens) > 0 THEN CAST(SUM(UniqueOpens) AS FLOAT)/CAST(SUM(Delivered) AS FLOAT) ELSE CAST(SUM(UniqueOpens) AS FLOAT) END as OpenRate
FROM [Non Triggered Sends Last Month]
GROUP BY EmailName

Answer 2

Your case statement needs to have a SUM in it too.

CASE WHEN UniqueOpens > 0

needs to be

CASE WHEN SUM(UniqueOpens) > 0

Answer 3

Why do you want a value when the unique delivered is non-positive? I would just go for NULL :

 SUM(UniqueOpens) * 1.0 / NULLIF(SUM(Delivered), 0) as OpenRate

However, if you want to prevent division by zero, I would suggest comparing the denominator , not the numerator:

 (CASE WHEN SUM(Delivered) > 0
       THEN SUM(UniqueOpens) * 1.0 / SUM(Delivered)
       ELSE 0  -- I'm assuming these SUM(Delivered) is a non-negative integer
  END) as OpenRate

I usually use the shorthand of * 1.0 to prevent integer division. Conversion to float is equally good.

Answer 4

Use sum Aggregation on Case statement

SELECT
EmailName,
SUM(Sent) as Sent,
SUM(Delivered) as Delivered,
SUM(UniqueOpens) as UniqueOpens,
SUM(TotalOpens) as TotalOpens,
SUM(UniqueClicks) as UniqueClicks,
SUM(TotalClicks) as TotalClicks,
SUM(Unsubscribes) as Unsubscribes,
SUM(CASE WHEN UniqueOpens > 0 THEN 
CAST(UniqueOpens AS FLOAT)/CAST(Delivered AS FLOAT)
 ELSE CAST(UniqueOpens AS FLOAT) )END as OpenRate
FROM [Non Triggered Sends Last Month]
GROUP BY EmailName