I have this situation:
select name,
subject
from Table_1
where date > getdate()-1
group by name, subject
order by id desc
union
select name,
subject
from table_2
where name not like 'abc%'
Table_1 and table_2 has similar structure.
I need to order by in SET1 UNION SET 2
This is not allowed in sql server. says "ORDER BY items must appear in the select list". I dont understand why the problem is. I am selecting equal number of columns on both queries. only that I want the result set together.
(on SQL Server 2017)
Anybody help!!
Thanks in advance.
Elaborating on my comment
select name,
subject
from Table_1
where date > getdate()-1
--group by name, subject --this isn't needed
union
select name,
subject
from table_2
where name not like 'abc%'
order by <yourCol> desc --notice change here
And for a conditional order by, there are a few posts on that.
Also, you don't need the group by since union removes duplicates.
But, the error is clear that the column you want to order by must be contained in the select list...
If you want to keep the first set ordered before the second set, just use a static column....
select name,
subject,
1 as Sort
from Table_1
where date > getdate()-1
--group by name, subject --this isn't needed
union
select name,
subject,
2 as Sort
from table_2
where name not like 'abc%'
order by Sort asc--notice change here
Related
I have a select query which returns a couple of rows grouped by ParentId. How can I add a new row with sum of a column after each parentId group?
For now I have kept the data in a temp table and the result is as below.
And I want to add a new row at the end of each ParentId group as below with the sum of column LoanAmount.
Any help will be appreciated. Many thanks.
You can use a common table expression to achieve this. Here I've created a cte with rank column for getting it sorted in order.
;WITH cte AS
(SELECT ParentId,
sum(LoanAmount) LoanAmount,
max(rank) + 1 AS rank
FROM test
GROUP BY ParentId)
SELECT *
FROM test
UNION ALL
SELECT *
FROM cte
ORDER BY ParentId, rank
rextester
See this link here enter link description here
I think you want:
SELECT ParentID, SUM(VALUE1), SUM(VALUE2)
FROM tableName
GROUP BY ID
You cant do it after each group or at the bottom like in excel, but you create a 'new table' in your query effectively.
Yeah having seen your updated comment, you main issue is youre thinking of it like excel, SQL is not a spreadsheet tool - its a relational database. Id suggest going through a SQL intro - youll pick up the concepts quite fast.
The query I gave you could be created as a stored procedure.
If you feel I've answered your question, id appreciate an upvote :)
You can make sum of group by subquery then combine them in union
; with cte as
( select 9999 as Slno, Level, ParentId, Ent_id, relation, sum(colname) as colname from table group by Level, ParentId, Ent_id, relation)
, ct as ( select row_number() over (partition by ParentId order by level) as Slno, Level, ParentId, Ent_id, Name, --- rest of your column names
colname from table
union all
select Slno, Level, ParentId, Ent_id, '' as Name, ---rest of '' for each column with column name as alias
colname from cte )
select Slno, Level, ParentId, Ent_Id, Name, ---- your columns of table
colname from ct order by Slno
This is just rough idea. Feel free to ask for any confusion.
Post your exact schema for accurate details.
I am trying to get available balance on last(max) date. I am trying to write below query but it is showing error.
select ACCOUNT_ID,AVAIL_BALANCE,OPEN_DATE,MAX(LAST_ACTIVITY_DATE)
from ACCOUNT
group by CUST_ID;
Column 'ACCOUNT.ACCOUNT_ID' is invalid in the select list because it
is not contained in either an aggregate function or the GROUP BY
clause.
I am new to sql. Can anyone let me know where I am wrong in this query?
Any column not having a calculation/function on it must be in the GROUP BY clause.
select ACCOUNT_ID,AVAIL_BALANCE,OPEN_DATE,MAX(LAST_ACTIVITY_DATE)
from ACCOUNT
group by ACCOUNT_ID,AVAIL_BALANCE,OPEN_DATE;
If you're wanting the most recent row for each customer, think ROW_NUMBER(), not GROUP BY:
;With Numbered as (
select *,ROW_NUMBER() OVER (
PARTITION BY CUST_ID
ORDER BY LAST_ACTIVITY_DATE desc) rn
from Account
)
select ACCOUNT_ID,AVAIL_BALANCE,OPEN_DATE,LAST_ACTIVITY_DATE
from Numbered
where rn=1
I think you want to select one records having max(LAST_ACTIVITY_DATE) for each CUST_ID.
For this you can use TOP 1 WITH TIES like following.
SELECT TOP 1 WITH TIES account_id,
avail_balance,
open_date,
last_activity_date
FROM account
ORDER BY Row_number()
OVER (
partition BY cust_id
ORDER BY last_activity_date DESC)
Issue with your query is, you can't select non aggregated column in select if you don't specify those columns in group by
If you want to get the max activity date for a customer then your query should be as below
select CUST_ID, MAX(LAST_ACTIVITY_DATE)
from ACCOUNT
group by CUST_ID;
You can't select any other column which is not in the group by clause. The error message also giving the same message.
with query(CUST_ID, LAST_ACTIVITY_DATE) as
(
select
CUST_ID,
MAX(LAST_ACTIVITY_DATE) as LAST_ACTIVITY_DATE
from ACCOUNT
group by CUST_ID
)
select
a.ACCOUNT_ID,
a.AVAIL_BALANCE,
a.OPEN_DATE,
a.LAST_ACTIVITY_DATE
from ACCOUNT as a
inner join query as q
on a.CUST_ID = q.CUST_ID
and a.LAST_ACTIVITY_DATE = q.LAST_ACTIVITY_DATE
I'm on SQL Server 2008, and having trouble querying an audit table the way I want to.
The table shows every time a new ID comes in, as well as every time an IDs Type changes
Record # ID Type Date
1 ae08k M 2017-01-02:12:03
2 liei0 A 2017-01-02:12:04
3 ae08k C 2017-01-02:13:05
4 we808 A 2017-01-03:20:05
I'd kinda like to produce a snapshot of the status for each ID, at a certain date. My thought was something like this:
SELECT
ID
,max(date) AS Max
FROM
Table
WHERE
Date < 'whatever-my-cutoff-date-is-here'
GROUP BY
ID
But that loses the Type column. If I add in the type column to my GROUP BY, then I'd get get duplicate rows per ID naturally, for all the types it had before the date.
So I was thinking of running a second version of the table (via a common table expression), and left joining that in to get the Type.
On my query above, all I have to join to are the ID & Date. Somehow if the dates are too close together, I end up with duplicate results (like say above, ae08k would show up once for each Type). That or I'm just super confused.
Basically all I ever do in SQL are left joins, group bys, and common table expressions (to then left join). What am I missing that I'd need in this situation...?
Use row_number()
select *
from ( select *
, row_number() over (partition by id order by date desc) as rn
from table
WHERE Date < 'whatever-my-cutoff-date-is-here'
) tt
where tt.rn = 1
I'd kinda like know how many IDs are of each type, at a certain date.
Well, for that you use COUNT and GROUP BY on Type:
SELECT Type, COUNT(ID)
FROM Table
WHERE Date < 'whatever-your-cutoff-date-is-here'
GROUP BY Type
Basing on your comment under Zohar Peled answer you probably looking for something like this:
; with cte as (select distinct ID from Table where Date < '$param')
select [data].*, [data2].[count]
from cte
cross apply
( select top 1 *
from Table
where Table.ID = cte.ID
and Table.Date < '$param'
order by Table.Date desc
) as [data]
cross apply
( select count(1) as [count]
from Table
where Table.ID = cte.ID
and Table.Date < '$param'
) as [data2]
I'm new to SQL in general and I need to delete all duplicates in a given database.
For the moment, I use this DB to experiment some things.
The table currently looks like this :
I know I can find all duplicates using this query :
SELECT COUNT(*) AS NBR_DOUBLES, Name, Owner
FROM dbo.animals
GROUP BY Name, Owner
HAVING COUNT(*) > 1
but I have a lot of trouble finding an adapted and updated solution to not only find all the duplicates, but also delete them all, only leaving one of each.
Thanks a lot for taking some of your time to help me.
;WITH numbered AS (
SELECT ROW_NUMBER() OVER(PARTITION BY Name, Owner ORDER BY Name, Owner) AS _dupe_num
FROM dbo.Animals
)
DELETE FROM numbered WHERE _dupe_num > 1;
This will delete all but one of each occurance with the same Name & Owner, if you need it to be more specific you should extend the PARTITION BY clause. If you want it to take in account the entire record you should add all your fields.
The record left behind is currently random, since it seems you do not have any field to have any sort of ordering on.
What you want to do is use a projection that numbers each record within a given duplicate set. You can do that with a Windowing Function, like this:
SELECT Name, Owner
,Row_Number() OVER ( PARTITION BY Name, Owner ORDER BY Name, Owner, Birth) AS RowNum
FROM dbo.animals
ORDER BY Name, Owner
This should give you results like this:
Name Owner RowNum
Ecstasy Sacha 1
Ecstasy Sacha 2
Ecstasy Sacha 3
Gremlin Max 1
Gremlin Max 2
Gremlin Max 3
Outch Max 1
Outch Max 2
Outch Max 3
Now you want to convert this to a DELETE statement that has a WHERE clause targeting rows with RowNum > 1. The way to use a windowing function with a DELETE is to first include the windowing function as part of a common table expression (CTE), like this:
WITH dupes AS
(
SELECT Name, Owner,
Row_Number() OVER ( PARTITION BY Name, Owner ORDER BY Name, Owner, Birth) AS RowNum
FROM dbo.animals
)
DELETE FROM dupes WHERE RowNum > 1;
This will delete later duplicates, but leave row #1 for each group intact. The only trick now is to make sure row #1 is the correct row, since not all of your duplicates have the same values for the Birth or Death columns. This is the reason I included the Birth column in the windowing function, while other answers (so far) have not. You need to decide if you want to keep the oldest animal or the youngest, and optionally change the Birth order in the OVER clause to match your needs.
Use CTE. I will show you a sample :
Create table #Table1(Field1 varchar(100));
Insert into #Table1 values
('a'),('b'),('f'),('g'),('a'),('b');
Select * from #Table1;
WITH CTE AS(
SELECT Field1,
RN = ROW_NUMBER()OVER(PARTITION BY Field1 ORDER BY Field1)
FROM #Table1
)
--SELECT * FROM CTE WHERE RN > 1
DELETE FROM CTE WHERE RN > 1
What I am doing is, numbering the rows. If there are duplicates based on PARTITION BY columns, it will be numbered sequentially, else 1.
Then delete those records whose count is greater than 1.
I won't spoon feed you solution hence you will have to play with PARTITION BY to reach your output
output :
Select * from #Table1;
Field1
---------
a
b
f
g
a
b
/*with cte as (...) SELECT * FROM CTE;*/
Field1 RN
------- -----
a 1
a 2
b 1
b 2
f 1
g 1
if NBR_DOUBLES had an ID field, I believe you could use this;
DELETE FROM NBR_DOUBLES WHERE ID IN
(
SELECT MAX(ID)
FROM dbo.animals
GROUP BY Name, Owner
HAVING COUNT(*) > 1
)
I have a table with data about cities that includes their name, population and other fields irrelevant to my question.
ID Name Population
1 A 45667
2 B 123456
3 C 3005
4 D 13769
To find the max population is basic, but I need a resulting table that has the max population in one column, and the corresponding city's name in another column
Population Name
123456 B
I've looked through similar questions, but for some reason the answers look over-complicated. Is there a way to write the query in 1 or 2 lines?
There are several ways that this can be done:
A filter in the WHERE clause:
select id, name, population
from yourtable
where population in (select max(population)
from yourtable)
Or a subquery:
select id, name, population
from yourtable t1
inner join
(
select max(population) MaxPop
from yourtable
) t2
on t1.population = t2.maxpop;
Or you can use TOP WITH TIES. If there can be no ties, then you can remove the with ties. This will include any rows that have the same population value:
select top 1 with ties id, name, population
from yourtable
order by population desc
Since you are using SQL Server you can also use ranking functions to get the result:
select id, name, population
from
(
select id, name, population,
row_number() over(order by population desc) rn
from yourtable
) src
where rn = 1
See SQL Fiddle with Demo of all.
As a side note on the ranking function, you might want to use dense_rank() instead of row_number(). Then in the event you have more than one city with the same population you will get both city names. (See Demo)