SQL query to compare subsets of rows between each other - sql

Question

There is an SQL query I'm after (in SQL Server). I need to get the count of instances where company Y is more expensive than company X. How would I start to tackle this? I've looked through various examples, but I cannot find anything similiar. I see PARTITION BY could be helpful, but not sure how to start from there - any hint will be very helpful.
ReadingId | Product | Price | Company
----------------------------------------------
1 | A | 3 | X
2 | A | 4 | Y
3 | A | 5 | Z
4 | B | 11 | X
5 | B | 12 | Y
6 | B | 13 | Z
...

Answer 1

One method is conditional aggregation. For each product:
select product,
max(case when company = 'Y' then price end) as Yprice
max(case when company = 'X' then price end) as Xprice
from t
group by product;
For a count, you can then do:
select count(*)
from (select product,
max(case when company = 'Y' then price end) as Yprice
max(case when company = 'X' then price end) as Xprice
from t
group by product;
) p
where Yprice > Xprice;
There are other methods as well. Pivot can be used, as well as a join with aggregation:
select count(*)
from t ty join
t tx
on ty.company = 'Y' and tx.company = 'X' and ty.product = tx.product
where ty.price > tx.price;
I should point out that all these methods sort of assume that X and Y only appear once for each product. That seems reasonable given your data.

Answer 2

Rather straight-forward. Get product prices for companies X and Y. Join them together on product and compare prices.
It assumes that each product is listed once for a company.
WITH
CTE_X
AS
(
SELECT Product, Price
FROM T
WHERE Company = 'X'
)
,CTE_Y
AS
(
SELECT Product, Price
FROM T
WHERE Company = 'Y'
)
SELECT COUNT(*) AS cc
FROM
CTE_X
INNER JOIN CTE_Y ON CTE_Y.Product = CTE_X.Product
WHERE
CTE_Y.Price > CTE_X.Price
;

Answer 3

You can do this with conditional aggregation.
with xandy as (select product,
max(case when company = 'X' then price end) as xprice,
max(case when company = 'Y' then price end) as yprice
from tablename
group by product)
select count(*)
from xandy
where yprice > xprice

Answer 4

This query, tho not efficient, will give you the details of what you need:
Select
CompanyY.*,
CompanyX.*
FROM
(
select * from OrderDetails
where Company = 'Y'
) CompanyY
JOIN
(
select * from OrderDetails
where Company = 'X'
) CompanyX
ON CompanyX.Product = CompanyY.Product
WHERE CompanyY.Price > CompanyX.Price
Try the SQLFiddle Here

Answer 5

You can use:
SELECT COUNT(*)
FROM (
SELECT Product
FROM mytable
GROUP BY Product
HAVING MAX(CASE WHEN Company = 'Y' THEN Price END)
>
MAX(CASE WHEN Company = 'X' THEN Price END) ) AS t
The sub-query returns the list of products where company Y is more expensive than company X. The outer query simply counts the number of these products.
Yet another version using a window function:
SELECT COUNT(*)
FROM (
SELECT Company,
ROW_NUMBER() OVER (PARTITION BY Product
ORDER BY Price DESC) AS rn
FROM mytable
WHERE Company IN ('X', 'Y')) AS t
WHERE t.rn = 1 AND Company = 'Y'
The sub-query filters out any rows not having either 'X' or 'Y' as their Company. The outer query counts the number of rows for which Company = 'Y' has the highest price.