Reputation: 914
Get a list of all customers order totals by month, and if the customer has no order in a given month, include a line for that month with 0 as the order total. In one statement? Totals already computed, no need for aggregate function.
Use of the coalesce function is acceptable.
Given list of customer order totals by month:
create table orders (cust char(1), month num, exps num);
insert into orders
values('a', 1, 5)
values('b', 2, 4)
values('c', 1, 8);
And a list of customers:
create table custs(cust char(1));
insert into custs
values('a')
values('b')
values('c')
values('d');
Generate this table:
cust, month, exps
a, 1, 5
a, 2, 0
b, 1, 0
b, 2, 4
c, 1, 8
c, 2, 0
d, 1, 0
d, 2, 0
Upvotes: 2
Views: 82
Reputation: 1396
select or1.cust, a.[month], sum(coalesce(or2.[exps], 0)) as exps
from (
select 1 as[month] union all select 2
) a cross join (select distinct cust from custs) or1
left join orders or2 on or2.[month] = a.[month] and or2.cust = or1.cust
group by or1.cust, a.[month]
order by or1.cust,a.[month]
And another version with picking up all existing months from the table. Results are same for our test data:
select or1.cust, a.[month], sum(coalesce(or2.[exps], 0)) as exps
from (
select distinct [month] from orders
) a cross join (select distinct cust from custs) or1
left join orders or2 on or2.[month] = a.[month] and or2.cust = or1.cust
group by or1.cust, a.[month]
order by or1.cust,a.[month]
Upvotes: 1
Reputation: 914
Making the cartesian product of customers and months was the first crack in the egg... and then a left join/coalesce w/ the result.
select all_possible_months.cust,
all_possible_months.month,
coalesce(orders.exps,0) as exps
from
(select order_months.month,
custs.cust
from
(select distinct month
from
orders
) as order_months,
custs
) all_possible_months
left join
orders on(
all_possible_months.cust = orders.cust and
all_possible_months.month = orders.month
);
Upvotes: 0