Gruppe am Tag

Question

Ich habe zwei Tabellen, Vertrieb und Anrufe wie folgt:

**Sales** 
CUST_ID   INT primary key, 
CUST_NM   Varchar(40), 
Sale_date   Datetime2, 
SALES    Money); 
CUST_ID CUST_NM   Sale_date    SALES 
1   Dom  2015-01-01 15:00:02.3000000 10.00 
2   Brian 2015-01-02 15:00:02.3000000 12.00 
3   Stu  2015-01-03 15:00:02.3000000 21.00 
4   John  2015-01-04 15:00:02.3000000 41.00 
5   Jack  2015-01-05 15:00:02.3000000 51.00 
6   Jill  2015-01-05 15:00:02.3000000 61.00 
7   Steve 2015-01-04 15:00:02.3000000 16.00 
8   Stacey 2015-01-03 15:00:02.3000000 19.00 
9   Lacey 2015-01-03 15:00:02.3000000 30.00 

fordert

NAME    Varchar(40), 
CALL_DATE   Date, 
TOTAL_CALLS  INT 


NAME  CALL_DATE TOTAL_CALLS 
Dom  2015-01-01 2 
Brian  2015-01-02 4 
Stu  2015-01-03 3 
John  2015-01-04 5 
Jack  2015-01-05 6 
Jill  2015-01-05 10 
Steve  2015-01-04 8 
Stacey  2015-01-03 7 
Lacey  2015-01-03 9 

ich eine select-Anweisung schreiben möchten, die das Datum, Brutto-Umsatz zurückbringt und die Gesamtanzahl der Anrufe von Sales und Calls am Tag des Beitritts.

Hier ist, was ich geschrieben habe, und ich denke, es sollte richtig sein, aber irgendwie bekomme ich nicht die richtige Ausgabe.

select Calls.CALL_DATE, sum(Sales.SALES) as gross_sale, sum(Calls.TOTAL_CALLS) as gross_total_calls 
from Sales 
join 
Calls 
on convert (date,sales.Sale_date)=calls.CALL_DATE 
group by Calls.CALL_DATE 
order by Calls.CALL_DATE 

Der Ausgang ich immer bin, ist

CALL_DATE gross_sale gross_total_calls 

2015-01-01 10.00  2 
2015-01-02 12.00  4 
2015-01-03 210.00  57 
2015-01-04 114.00  26 
2015-01-05 224.00  32 

Wohin gehe ich falsch ??

Answer 1

Sie generieren für jeden Tag ein kartesisches Produkt. Sie müssen vor dem join aggregieren. Oder Sie können dies tun, mit einer union all und Aggregation:

select dte, sum(sales) as sales, sum(calls) as total_calls 
from ((select cast(s.sale_date as date) as dte, sales, 0 as calls 
     from sales s 
    ) union all 
     (select call_date, 0, total_calls as calls 
     from calls c 
    ) 
    ) sc 
group by dte 
order by dte; 

Answer 2

Der alternative Weg wie von Gordon vorgeschlagen Variablen Tabelle ein lauffähiges Testskript zu erstellen. Beachten Sie die zwei zusätzlichen Datenzeilen und den FULL OUTER JOIN, mit dem alle Daten zurückgegeben werden können.

declare @Sales table (CUST_ID INT primary key, CUST_NM Varchar(40), Sale_date Datetime2,SALES Money); 
insert into @Sales (CUST_ID, CUST_NM, Sale_date, SALES) 
select 1, 'Dom', '2015-01-01 15:00:02.3000000', 10.00 union 
select 2, 'Brian', '2015-01-02 15:00:02.3000000', 12.00 union 
select 3, 'Stu', '2015-01-03 15:00:02.3000000', 21.00 union 
select 4, 'John', '2015-01-04 15:00:02.3000000', 41.00 union 
select 5, 'Jack', '2015-01-05 15:00:02.3000000', 51.00 union 
select 6, 'Jill', '2015-01-05 15:00:02.3000000', 61.00 union 
select 7, 'Steve', '2015-01-04 15:00:02.3000000', 16.00 union 
select 8, 'Stacey', '2015-01-03 15:00:02.3000000', 19.00 union 
select 9, 'Lacey', '2015-01-03 15:00:02.3000000', 30.00 union 
select 10, 'Tom', '2015-01-07 15:00:02.3000000', 1.00 

declare @Calls table (NAME Varchar(40), CALL_DATE Date, TOTAL_CALLS INT) 
insert into @Calls (NAME, CALL_DATE, TOTAL_CALLS) 
select 'Dom', '2015-01-01', 2 union 
select 'Brian', '2015-01-02', 4 union 
select 'Stu', '2015-01-03', 3 union 
select 'John', '2015-01-04', 5 union 
select 'Jack', '2015-01-05', 6 union 
select 'Jill', '2015-01-05', 10 union 
select 'Steve', '2015-01-04', 8 union 
select 'Stacey', '2015-01-03', 7 union 
select 'Lacey', '2015-01-03', 9 union 
select 'Tom', '2015-01-06', 1 

select * from @Sales 
select * from @Calls 

select ISNULL (a.CALL_DATE, b.CALL_DATE) as CALL_DATE, gross_sale, TOTAL_CALLS 
from 
(select convert(date, Sale_date) as CALL_DATE, sum(SALES) as gross_sale 
    from @Sales 
    group by convert(date, Sale_date) 
) a 
full outer join 
(select CALL_DATE, SUM(TOTAL_CALLS) as TOTAL_CALLS 
    from @Calls 
    group by CALL_DATE 
) b on a.CALL_DATE = b.CALL_DATE 
order by a.CALL_DATE