SQL say that there are no rows with a certain value - sql-server

Question

I'm new with SQL and I need your help.
I have a table with a Name column and a Value column.
This value represents a 3-values enum, let's say Yes, No, and Maybe.
I'm trying to count, for each name, how many times I have each value of the enum. Something like this:
Name1 Yes 3
Name1 No 1
Name1 Maybe 4
Name2 Yes 0
Name2 No 1
Name2 Maybe 2
Name3 Yes 1
Name3 No 0
Name3 Maybe 7
but there are times when I don't have any entry for a given name with a given value, and it doesn't show up. Shortly, when count is 0, I don't get the row.
And I have to put it in a stored procedure.
How can I achieve this?
Thank you all in advance!
P.S. I'm using SQL Server
EDIT:
What I tried is:
SELECT Name,
Value,
COUNT(Value) as ValueCount
from Data
group by Name, Value
order by Name

Answer 1

Use a cross join to generate all the rows. Then use a left join to join in the values for counting:
SELECT n.Name, v.value, COUNT(d.value) as ValueCount
from (select distinct name from Data) n cross join
(select distinct value from Data) v left join
data d
on d.name = n.name and d.value = v.value
group by n.Name, v.Value
order by n.Name;
You don't specify the database. This is standard SQL and should work in any database.

Answer 2

You can use two cte to get a list of all combinations you want to show, then do a left join with your original table.
WITH enum AS
(
SELECT 'Yes' AS Value
UNION
SELECT 'No'
UNION
SELECT 'Maybe'
), combs AS
(
SELECT DISTINCT d.Name, e.Value
FROM enum e, Data d
)
SELECT c.Name, c.Value, COUNT(d.Name) FROM combs c
LEFT JOIN data d ON d.Name = c.Name AND d.Value = c.Value
GROUP BY c.Name, c.Value
ORDER BY c.Name, c.Value

Answer 3

Looking at your table overall, it would be a lot cleaner like this:
Name Yes No Maybe
__________________________
Bob 3 4 0
Tim 0 1 3
Jim 2 7 2
I know that's not really an answer to solving your problems but this table is much smaller and achieves the same results.

Answer 4

Often when trying to count something that isn't there, it's best to join to a driver table, in this case, you need all name/value pairs to join against, created below in a simple cte using a CROSS JOIN, but you could pull the values in a number of ways:
;with cte AS (SELECT DISTINCT d1.name,d2.value
FROM data d1
CROSS JOIN data d2
)
SELECT cte.Name,
cte.Value,
COALESCE(COUNT(d.Value),0) as ValueCount
from cte
LEFT JOIN Data d
ON d.value = cte.value
AND d.name = cte.name
group by cte.Name, cte.Value
order by cte.Name
Since this requires a LEFT JOIN any missing values will return NULL, so COALESCE() is used to show 0.