Consider a table like this
table
+--------+---------+-----------+---------+-----------+
| BookId | ItemId1 | Quantity1 | ItemId2 | Quantity2 |
+--------+---------+-----------+---------+-----------+
| 1 | 1 | 2 | 2 | 1 |
| 1 | 3 | 1 | 2 | 1 |
| 2 | 1 | 1 | 2 | 1 |
+--------+---------+-----------+---------+-----------+
Now I want to get the sum of columns quantity for each item grouped by book. How can I take the sum across different columns then? right now I use an awkward solution like building a temporary table and then querying this one, but it must be possible in a more elegant way!?
select
BookId 'BookId',
ItemId1 'ItemId',
Quantity1 'Quantity'
into #temptab
from table
union all
select
BookId,
ItemId2,
Quantity2
from table
and after that
select
BookId,
ItemId,
sum(Quantity)
from #temptab
group by ItemId, BookId
How can I get rid of this intermediate step?
Desired output:
+--------+--------+----------+
| BookId | ItemId | Quantity |
+--------+--------+----------+
| 1 | 1 | 2 |
| 1 | 3 | 1 |
| 1 | 2 | 2 |
| 2 | 1 | 1 |
| 2 | 2 | 1 |
+--------+--------+----------+
Use cross apply with table valued constructor to unpivot the data then find sum per bookid and item.
This will avoid your intermediate step
SELECT BookId,
item,
Sum(quantity)
FROM Youratble
CROSS apply (VALUES(Quantity1,ItemId1),
(Quantity2,ItemId2))cs(quantity, item)
GROUP BY BookId,
item
As mentioned by Mikael Eriksson for sql server 2005 use this
SELECT BookId,
item,
Sum(quantity)
FROM Youratble
cross apply
(select Quantity1, ItemId1
union all
select Quantity2, ItemId2) as cs(quantity, item)
GROUP BY BookId,
item
Based on what you've done, you can do it in one single step using subqueries:
select
combined.BookId,
combined.ItemId,
sum(combined.Quantity)
from
(select
BookId 'BookId',
ItemId1 'ItemId',
Quantity1 'Quantity'
from tableName
union all
select
BookId,
ItemId2,
Quantity2
from tableName) as combined
group by combined.ItemId, combined.BookId
Related
I have a table showing simple order details as follows
+---------+---------+---------+
| Order_id| Item_id |Quantity |
+---------+---------+---------+
| 10001 | 1 | 4 |
+---------+---------+---------+
Now I need to show each quantity as a line item as follows.
+---------+---------+---------+---------+
| Order_id| Item_id | Line_id |Quantity |
+---------+---------+---------+---------+
| 10001 | 1 | 1 | 1 |
+---------+---------+---------+---------+
| 10001 | 1 | 2 | 1 |
+---------+---------+---------+---------+
| 10001 | 1 | 3 | 1 |
+---------+---------+---------+---------+
| 10001 | 1 | 4 | 1 |
+---------+---------+---------+---------+
Can any one help me how can I write query to achieve this?
We can try using a recursive CTE here:
WITH cte AS (Order_id, Item_id, Line_id, Amount) AS (
SELECT Order_id, Item_id, Quantity, 1
FROM yourTable
UNION ALL
SELECT Order_id, Item_id, Line_id - 1, 1
FROM cte
WHERE Line_id - 1 >= 1
)
SELECT Order_id, Item_id, Line_id, Quantity
FROM cte
ORDER BY Order_id, Item_id, Line_id;
The logic in the CTE above is that we form the base case of the recursion using the following tuple from your source table:
(10001, 1, 4, 1)
That is, we start counting for the Line_id from 4, decreasing by one in each step of the recursion. At the end of the recursive case, we hit a Line_id value of 1.
You can use unnest on an array to get the value you want:
select od.order_id, od.item_id, u.line_number, u.quantity
from order_details od cross join
unnest(repeat(1, od.quantity)) with ordinality u(quantity, line_number)
I have some queries in SQL Server. I have two tables
keyword_text
Keyword_relate
Columns in keyword_text:
key_id
keywords
Columns in keyword_relate:
key_id
product_id
score
status
Sample data for keyword_text:
----|----------
1 | Pencil
2 | Pen
3 | Books
Sample data for keyword_relate:
----------------------------
Sno| Product | SCore|status
---------------------------
1 | 124 | 2 | 1
1 | 125 | 3 | 1
2 | 124 | 3 | 1
2 | 125 | 2 | 1
From this I want to get the product_id, grouped by keywords and which have maximum score
Presuming that key_id of first table is Sno in second table. You can use ROW_NUMBER:
WITH CTE AS
(
SELECT Product AS ProductID, Score As MaxScore,
RN = ROW_NUMBER() OVER (PARTITION BY kt.key_id ORDER BY Score DESC)
FROM keyword_text kt INNER JOIN keyword_relate kr
ON kt.key_id = kr.Sno
)
SELECT ProductID, MaxScore
FROM CTE
WHERE RN = 1
I have written a query which selects lets say 10 rows for this example.
+-----------+------------+
| STORENAME | COMPLAINTS |
+-----------+------------+
| Store1 | 4 |
| Store7 | 2 |
| Store8 | 1 |
| Store9 | 1 |
| Store2 | 1 |
| Store3 | 1 |
| Store4 | 1 |
| Store5 | 0 |
| Store6 | 0 |
| Store10 | 0 |
+-----------+------------+
How would I go about displaying the TOP 3 rows BUT Having the remaining rows roll up into a row called "other", and it adds all of their Complaints together?
So like this for example:
+-----------+------------+
| STORENAME | COMPLAINTS |
+-----------+------------+
| Store1 | 4 |
| Store7 | 2 |
| Store8 | 1 |
| Other | 4 |
+-----------+------------+
So what has happened above, is it displays the top3 then adds the complaints of the remaining rows into a row called other
I have exhausted all my resources and cannot find a solution. Please let me know if this makes sense.
I have created a SQLfiddle of the above tables that you can edit if it is possible :)
Here's hoping this is possible :)
Thanks,
Mike
Something like this may work
select *, row_number() over (order by complaints desc) as sno
into #temp
from
(
SELECT
a.StoreName
,COUNT(b.StoreID) AS [Complaints]
FROM Stores a
LEFT JOIN
(
SELECT
StoreName
,Complaint
,StoreID
FROM Complaints
WHERE Complaint = 'yes') b on b.StoreID = a.StoreID
GROUP BY a.StoreName
) as t ORDER BY [Complaints] DESC
select storename,complaints from #temp where sno<4
union all
select 'other',sum(complaints) as complaints from #temp where sno>=4
I do this with double aggregation and row_number():
select (case when seqnum <= 3 then storename else 'Other' end) as StoreName,
sum(numcomplaints) as numcomplaints
from (select c.storename, count(*) as numcomplaints,
row_number() over (order by count(*) desc) as seqnum
from complaints c
where c.complaint = 'Yes'
group by c.storename
) s
group by (case when seqnum <= 3 then storename else 'Other' end) ;
From what I can see, you don't really need any additional information from stores, so this version just leaves that table out.
Given a table A has the following data:
+----------+-------+
| Supplier | buyer |
+----------+-------+
| A | 1 |
| A | 2 |
| B | 3 |
| B | 4 |
| B | 5 |
+----------+-------+
My question is, can I transpose the second column so the resultant table will be like:
+----------+--------+--------+--------+
| Supplier | buyer1 | buyer2 | buyer3 |
+----------+--------+--------+--------+
| A | 1 | 2 | |
| B | 3 | 4 | 5 |
+----------+--------+--------+--------+
Assuming the maximum number of buyers is known as three.
You could use a common table expression to give each buyer an order within the supplier, and then just do a regular case to put them in columns;
WITH cte AS (
SELECT supplier, buyer,
ROW_NUMBER() OVER (PARTITION BY supplier ORDER BY buyer) rn
FROM Table1
)
SELECT supplier,
MAX(CASE WHEN rn=1 THEN buyer END) buyer1,
MAX(CASE WHEN rn=2 THEN buyer END) buyer2,
MAX(CASE WHEN rn=3 THEN buyer END) buyer3
FROM cte
GROUP BY supplier;
An SQLfiddle to test with.
You may consider using PIVOT clause:
select *
from (
select supplier, buyer, row_number() over (partition by supplier order by buyer) as seq
from a
)
pivot (max(buyer) for seq in (1 as buyer1, 2 as buyer2, 3 as buyer3));
SQLFiddle here.
We have a table like this:
+----+--------+
| Id | ItemId |
+----+--------+
| 1 | 1100 |
| 1 | 1101 |
| 1 | 1102 |
| 2 | 2001 |
| 2 | 2002 |
| 3 | 1101 |
+----+--------+
We want to count how many items each guy has, and show the guys with 2 items or more. Like this:
+----+-----------+
| Id | ItemCount |
+----+-----------+
| 1 | 3 |
| 2 | 2 |
+----+-----------+
We didn't count the guy with Id = 3 because he's got only 1 item.
How can we do this in SQL?
SELECT id, COUNT(itemId) AS ItemCount
FROM YourTable
GROUP BY id
HAVING COUNT(itemId) > 1
Use this query
SELECT *
FROM (
SELECT COUNT(ItemId ) AS COUNT, Id FROM ITEM
GROUP BY Id
)
my_select
WHERE COUNT>1
SELECT id,
count(1)
FROM YOUR_TABLE
GROUP BY id
HAVING count(1) > 1;
select Id, count(ItemId) as ItemCount
from table_name
group by Id
having ItemCount > 1