具有非唯一日期的SQL Sum

Question

I'm trying to write a SQL query that will sum total production from the following two example tables:

Table: CaseLots
DateProduced        kgProduced
October 1, 2013     10000
October 1, 2013     10000
October 2, 2013     10000

Table: Budget
OperatingDate       BudgetHours
October 1, 2013     24
October 2, 2013     24

I would like to output a table as follows:

TotalProduction   TotalBudgetHours    
30000                   48

Here is what I have for code so far:

SELECT 
Sum(kgProduced) AS TotalProduction, Sum(BudgetHours) AS TotalBudgetHours
FROM 
dbo.CaseLots INNER JOIN dbo.Budget ON dbo.CaseLots.DateProduced = dbo.Budget.OperatingDate
WHERE 
dbo.Budget.OperatingDate BETWEEN '2013-10-01' AND '2013-10-02'

It seems that the query is double summing the budget hour in instances where more than one case lot is produced in a day. The table I'm getting is as follows:

Total Production     BudgetHours
30000                72

How do I fix this?

Answer 1

Think about what the INNER JOIN is doing.

For every row in CaseLot, its finding any row in Budget that has a matching date.

If you were to remove your aggregation statements in SQL, and just show the inner join, you would see the following result set:

DateProduced kgProduced OperatingDate BudgetHours

October 1, 2013 10000 October 1, 2013 24

October 1, 2013 10000 October 1, 2013 24

October 2, 2013 10000 October 2, 2013 24

(dammit StackOverflow, why don't you have Markdown for tables :( )

Running your aggregation on top of that it is easy to see how you get the 72 hours in your result.

The correct query needs to aggregate the CaseLots table first, then join onto the Budget table.

SELECT DateProduced, TotalKgProduced, SUM(BudgetHours) AS TotalBudgetHours
FROM
(
    SELECT DateProduced, SUM(kgProduced) AS TotalKgProduced
    FROM CaseLots
    GROUP BY DateProduced
) AS TotalKgProducedByDay
INNER JOIN
Budget
ON TotalKgProducedByDay.DateProduced = Budget.OperatingDate
WHERE DateProduced BETWEEN '1 Oct 2013' AND '2 Oct 2013'
GROUP BY DateProduced

Answer 2

The problem is in the INNER JOIN produces a 3 row table since the keys match on all. So there is three '24's with a sum of 72.

To fix this, it would probably be easier to split this into two queries.

SELECT Sum(kgProduced) AS TotalProduction
FROM dbo.CaseLots
WHERE dbo.CaseLots.OperatingDate BETWEEN '2013-10-01' AND '2013-10-02'

LEFT JOIN

SELECT Sum(BudgetHours) AS TotalBudgetHours
FROM dbo.Budget
WHERE dbo.Budget.OperatingDate BETWEEN '2013-10-01' AND '2013-10-02'

Answer 3

This could be easily achieved by this:

SELECT
   (SELECT SUM(kgProduced) FROM dbo.CaseLots WHERE DateProduced BETWEEN '2013-10-01' AND '2013-10-02') AS TotalProduction,
   (SELECT SUM(BudgetHours) FROM dbo.Budget WHERE OperatingDate BETWEEN '2013-10-01' AND '2013-10-02') AS TotalBudgetHours

There's no need for joining the two tables.

Answer 4

Try this:

select DateProduced,TotalProduction,TotalBudgetHours from
  (select DateProduced,sum(kgProduced) as TotalProduction 
  from CaseLots group by DateProduced) p
  join
  (select OperatingDate,sum(BudgetHours) as TotalBudgetHours 
  from Budget group by OperatingDate) b
  on (p.DateProduced=b.OperatingDate)
where p.DateProduced between '2013-10-01' AND '2013-10-02'

Answer 5

The other answers are simpler for this particular case. However if you needed to SUM 10 different values on the CaseLots table, you'd need 10 different subqueries. The following is a general, more scaleable solution:

SELECT 
    SUM(DayKgProduced) AS TotalProduction, 
    SUM(BudgetHours) AS TotalBudgetHours
FROM (
    SELECT 
        DateProduced,
        SUM(kgProduced) AS DayKgProduced, 
    FROM dbo.CaseLots 
    WHERE DateProduced BETWEEN '2013-10-01' AND '2013-10-02' 
    GROUP BY DateProduced
) DailyTotals
INNER JOIN dbo.Budget b ON DailyTotals.DateProduced = b.OperatingDate

First you SUM the production of each CaseLot without having to SUM the BudgetHours. If you used a SELECT * FROM in the query above you'd see:

Date        DayKgProduced  BudgetHours
2013-10-01  20000          24 
2013-10-02  10000          24

But you want the overall total, so we SUM those daily values, correctly producing:

TotalProduction  TotalBudgetHours
30000            48