Question

I have three tables:

Users
    UserId
    CategoryId
Categories
    CategoryId
    CategoryName
Addresses
    AddressId
    UserId
    AddressValue

Users have to be in a category, but they don’t have to have an address. They can have more than one address.

I want to return the number of users in each category who have at least 1 address. So, I need to return something like CategoryId,NumberOfUsers,NumberOfUsersWithAddresses, where the following condition holds:

NumberOfUsers >= NumberOfUsersWithAddresses

I have this to return the number of users in each category:

SELECT
    Categories.CategoryId, COUNT(Users.NumberOfUsers) AS Occurances
FROM
    Users INNER JOIN
    Categories on Users.CategoryId = Categories.CategoryId
GROUP BY
    Categories.CategoryId

I tried this:

SELECT
    Categories.CategoryId, COUNT(Users.NumberOfUsers) AS Occurances
    (SELECT Count(Users.UserId)
        FROM Addresses INNER JOIN Users on Addresses.UserId = Users.UserId
        WHERE Users.CategoryId = Category.CategoryID
        GROUP BY Users.CategoryID) AS NumberOfUsersWithAddresses
FROM
    Users INNER JOIN
    Categories on Users.CategoryId = Categories.CategoryId
GROUP BY
    Categories.CategoryId

But this didn’t work – it returns the number of addresses in each category, rather than the number of users with addresses in each category. So if a user has two addresses, she counts as a ‘2’ rather than a ‘1’ in the sum.

What query do I need to make this work?

Answer 1

Not too fancy but a subquery in the SELECT clause might do it.
AFAIK, the COUNT(DISTINCT u.UserID) was all you missed.

SELECT  CategoryName
        , NumberOfUsers = (
            SELECT  COUNT(*) 
            FROM    Users 
            WHERE   CategoryId = c.CategoryId
          ) 
        , NumberOfUsersWithAddresses = (
            SELECT  COUNT(DISTINCT u.UserID) 
            FROM    Users u 
                    INNER JOIN Addresses a ON a.UserID = u.UserID 
            WHERE   CategoryId = c.CategoryId
          ) 
FROM    Categories c
GROUP BY
        CategoryName

or

SELECT  CategoryName
        , NumberOfUsers = COUNT(*)
        , NumberOfUsersWithAddresses = (
            SELECT  COUNT(DISTINCT u.UserID) 
            FROM    Users u 
                    INNER JOIN Addresses a ON a.UserID = u.UserID 
            WHERE   CategoryId = c.CategoryId
          ) 
FROM    Categories c
        INNER JOIN Users u ON u.CategoryID = c.CategoryID
GROUP BY
        CategoryName