Select the newest record - sql

I would like to run a select statement that runs and select only the newest record by Recored_timestampe field for the keys teacher_id and student_id. So any time, it runs it needs to provide only one record. how could I do it, please? The output could be without the field Recored_timestampe. Thanks

Using the window function,partitioned by teacher_id and student_id and sorting it by recorded_timestamp will give you the desired result.
select * from(select teacher_id,student_id,teacher_name,comment ,recorded_timestamp, row_number() over(partition by teacher_id,student_id order by recorded_timestamp desc)as rownum from temp0607)out1 where rownum=1
Also you may have to look at the way recorded_timestamp is stored. If it is stored as string, you can convert it into timestamp using from_unixtime(unix_timestamp(recorded_timestamp,'dd/MM/yyyy HH:mm'),'dd/MM/yyyy HH:mm')

First, arrange the record by datetime
SELECT *,RANK() OVER (PARTITION BY student_id ORDER BY Recored_timestamp desc) as ranking
FROM #temp
Then, if you want to know the newest record with student_id which is not null, then you can use OUTER APPLY to add a column which is non-NULL student_id.
OUTER APPLY (SELECT student_id
FROM #temp
WHERE #temp.teacher_id = ranktable.teacher_id
AND student_id IS NOT NULL
) AS jointable
Here is an example:
Create Table #temp
(
teacher_id int
,student_id int
,teacher_name varchar(40)
,comment varchar(100)
,Recored_timestamp datetime
)
INSERT INTO #temp
VALUES
(449,111,'lucy','Could be better','2021-05-04 07:00:00.000')
,(449,null,'lucy','smily','2021-05-11 07:00:00.000')
,(449,111,'lucy','not listening','2021-05-08 07:00:00.000')
,(448,null,'Toni','Good','2021-06-04 09:00:00.000')
,(448,222,'Toni','not doing as expected','2021-06-04 08:00:00.000')
SELECT DISTINCT teacher_id,
jointable.student_id,
teacher_name,
comment,
Recored_timestamp,
ranking
FROM
(
SELECT *,RANK() OVER (PARTITION BY teacher_id ORDER BY Recored_timestamp DESC) AS ranking
FROM #temp
) AS ranktable
OUTER APPLY (SELECT student_id
FROM #temp
WHERE #temp.teacher_id = ranktable.teacher_id
AND student_id IS NOT NULL
) AS jointable
WHERE ranking = 1 --only newest record will be extracted
Drop table #temp

You can base from this query to get the newest data.
SELECT TOP 1 * FROM tablename T1
INNER JOIN(SELECT teacher_id, Max(Recored_timestamp) as MaxDate from tablename GROUP BY teacher_id) T2 ON T2.teacher_id = T1.teacher_id AND T1.Recored_timestamp = T2.MaxDate

Related

Turn these temp tables into one longer subquery (can't use Temp tables in Power BI)

Currently I have created these temp tables to get the desired output I need. However, Power BI doesn't allow the use of temp tables so I need to get this all into 1 query using inner selects.
drop table if exists #RowNumber
Select Date, ID, ListID
, row_number() over (partition by ID order by ID) as rownum
into #RowNumber
from Table
where Date= cast(getdate()-1 as date)
group by Date, ID, ListID
order by ID
drop table if exists #1stListIDs
select ListID as FirstID, ID, Date
into #1stListIDs
from #RowNumber
where rownum = 1
drop table if exists #2ndlistids
Select ListID as SecondListID, ID, Date
into #2ndlistids
from #RowNumber
where rownum = 2
--Joins the Two Tables back together to allow the listids to be in the same row
drop table if exists #FinalTableWithTwoListIDs
select b.FirstListID, a.SecondListID, a.ID, a.Date
into #FinalTableWithTwoListIDs
from #2ndlistids a
join #1stListIDs b on a.ID= b.ID
order by ID
This code is simple and straight forward. However I can't seem to figure out using a subquery. Here is what I have. It works for the FirstListID select statement, but not the SecondListID portion. I believe this is because you can't reference the inner most select statement with multiple different outer select statements, but I could be wrong.
Select a.ListId as SecondListID, a.ID, a.Date
from (
select a.ListId as FirstListID, a.ID, a.Date
from (
Select Date, ID, ListId
, row_number() over (partition by ID order by ID) as rownum
from Table
where Date = cast(getdate()-1 as date)
group by Date, ID, ListId
order by ID) a
where a.rownum = 1) b
where a.rownum = 2) c
Just to show, for completeness, how you could use CTEs to replace the #temp tables, it would be something along the lines of
with RowNumber as (
select Date, ID, ListID
, row_number() over (partition by ID order by ID) as rownum
from Table
where Date= cast(dateadd(day,-1,getdate()) as date)
group by Date, ID, ListID
),
FirstListIDs as (
select ListID as FirstID, ID, Date
from RowNumber
where rownum = 1
),
SecondListIDs as (
select ListID as SecondID, ID, Date
from RowNumber
where rownum = 2
)
select f.FirstID, s.SecondID, s.ID, s.Date
from Secondlistids s
join FirstListIDs f on s.ID=f.ID
order by s.ID
Note the use of dateadd which is recommended over the ambiguousdate +/- value assumed to be days, and where relevant meaningful table aliases.
You could do it with a CTE and joining the two together, but that is inefficient and unnecessary.
It looks like you just need LAG to get the previous ListID
I note that PARTITION BY ID ORDER BY ID is non-deterministic and the ordering will be random. I strongly suggest you find a deterministic ordering.
SELECT
PrevID AS b.FirstListID,
ListID AS a.SecondListID,
ID,
Date
FROM (
SELECT
Date,
ID,
ListID,
ROW_NUMBER() OVER (PARTITION BY ID ORDER BY ID) AS rownum,
LAG(ListID) OVER (PARTITION BY ID ORDER BY ID) AS PrevID
from [Table]
where Date = cast(getdate() - 1 as date)
group by Date, ID, ListID
) AS WithRowAndLag
WHERE rownum = 2;
ORDER BY ID;

SQL - All values from first table joined with first value from second one and null otherwise

is it possible to do a join like this in SQL:
We want to join the two tables on SessionID, but the resulting table should return full table "Pageviews" combined with only a single value of Sessions for each unique SessionID (for others the value should be NULL, please see the 'Desired Result' table for explanation).
Thank you.
Based on you requirement( and truly based on the data provided) the below code would do the work -
declare #pageviews table(sessionid char(10), pageviews tinyint)
declare #sessions table(sessionid char(10), sessions tinyint)
insert into #pageviews values
('FA-1',34),
('FA-1',36),
('FA-2',23),
('FA-3',11),
('FA-3',32),
('FA-3',25)
insert into #sessions values
('FA-1',23),
('FA-2',14),
('FA-3',9)
;with cte as
(select sessionid,pageviews,ROW_NUMBER() OVER(PARTITION BY sessionid ORDER BY sessionid desc) rn
from #pageviews)
select p.sessionid,p.pageviews,case when rn = 1 then s.sessions else null end as sessions
from cte p inner join #sessions s
on p.sessionid = s.sessionid
You can do like this.
SELECT T2.SessionID,T2.pageviews,T1.Sessions
FROM sessions AS T1
RIGHT JOIN
(
SELECT ROW_NUMBER() OVER(PARTITION BY sessionid ORDER BY sessionid DESC) AS ROWNUM,
sessionid,
pageviews
FROM pageviews
) AS T2
ON T1.sessionid = T2.sessionid AND T2.ROWNUM = 1
SQLFiddle

Sql select distinct row by a columns highest value

I am having an issue trying to select one row per city name. This is the following collection I am getting:
This is my query so far:
select pl.PlaceId,
pl.Name,
pop.NumberOfPeople,
pop.Year
from dbo.Places pl
inner join dbo.Populations pop
on pop.PlaceId = pl.PlaceId
where pop.NumberOfPeople >= 1000
and pop.NumberOfPeople <= 99999
I am trying to get it to where it only selects a city one time, but uses the most recent date. So in the above picture, I would only see Abbeville for 2016 and not 2015. I believe I need to do either a group by or do a sub query to flatten the results. If anybody has any advice on how I can handle this, it will be greatly appreciated.
Assuming you are using SQLSERVER,you can use Rownumber
;with cte
as
(select pl.PlaceId,
pl.Name,
pop.NumberOfPeople,
pop.Year,
row_number() over(partition by pl.Name order by year desc) as rownum
from dbo.Places pl
inner join dbo.Populations pop
on pop.PlaceId = pl.PlaceId
where pop.NumberOfPeople >= 1000
and pop.NumberOfPeople <= 99999
)
select * from cte where rownum=1
The following query serves the purpose.
CREATE TABLE #TEMP_TEST
(
PlaceId INT,
Name VARCHAR(50),
NumberOfPeople INT,
YEAR INT
)
INSERT INTO #TEMP_TEST
SELECT 1,'Abbeville',2603,2016
UNION
SELECT 5,'Alabester',32948,2016
UNION
SELECT 9,'Aubum',63118,2016
UNION
SELECT 1,'Abbeville',2402,2015
UNION
SELECT 5,'Alabester',67902,2017
SELECT PlaceId, Name, NumberOfPeople, YEAR FROM
(
SELECT ROW_NUMBER() OVER (PARTITION BY PlaceId ORDER BY YEAR DESC) RNO,
PlaceId, Name, NumberOfPeople, YEAR
FROM #TEMP_TEST
)T
WHERE RNO = 1
DROP TABLE #TEMP_TEST

how to get the distinct records based on maximum date?

I'm working with Sql server 2008.i have a table contains following columns,
Id,
Name,
Date
this table contains more than one record for same id.i want to get distinct id having maximum date.how can i write sql query for this?
Use the ROW_NUMBER() function and PARTITION BY clause. Something like this:
SELECT Id, Name, Date FROM (
SELECT *, ROW_NUMBER() OVER (PARTITION BY Id ORDER BY Date desc) AS ROWNUM
FROM [MyTable]
) x WHERE ROWNUM = 1
If you need only ID column and other columns are NOT required, then you don't need to go with ROW_NUMBER or MAX or anything else. You just do a Group By over ID column, because whatever the maximum date is you will get same ID.
SELECT ID FROM table GROUP BY ID
--OR
SELECT DISTINCT ID FROM table
If you need ID and Date columns with maximum date, then simply do a Group By on ID column and select the Max Date.
SELECT ID, Max(Date) AS Date
FROM table
GROUP BY ID
If you need all the columns but 1 line having Max. date then you can go with ROW_NUMBER or MAX as mentioned in other answers.
SELECT *
FROM table AS M
WHERE Exists(
SELECT 1
FROM table
WHERE ID = M.ID
HAVING M.Date = Max(Date)
)
One way, using ROW_NUMBER:
With CTE As
(
SELECT Id, Name, Date, Rn = Row_Number() Over (Partition By Id
Order By Date DESC)
FROM dbo.TableName
)
SELECT Id --, Name, Date
FROM CTE
WHERE Rn = 1
If multiple max-dates are possible and you want all you could use DENSE_RANK instead.
Here's an overview of sql-server's ranking function: http://technet.microsoft.com/en-us/library/ms189798.aspx
By the way, CTE is a common-table-expression which is similar to a named sub-query. I'm using it to be able to filter by the row_number. This approach allows to select all columns if you want.
select Max(Date) as "Max Date"
from table
group by Id
order by Id
Try with Max(Date) and GROUP BY the other two columns (the ones with repeating data)..
SELECT ID, Max(Date) as date, Name
FROM YourTable
GROUP BY ID, Name
You may try with this
DECLARE #T TABLE(ID INT, NAME VARCHAR(50),DATE DATETIME)
INSERT INTO #T VALUES(1,'A','2014-04-20'),(1,'A','2014-04-28')
,(2,'A2','2014-04-22'),(2,'A2','2014-04-24')
,(3,'A3','2014-04-20'),(3,'A3','2014-04-28')
,(4,'A4','2014-04-28'),(4,'A4','2014-04-28')
,(5,'A5','2014-04-28'),(5,'A5','2014-04-28')
SELECT T.ID FROM #T T
WHERE T.DATE=(SELECT MAX(A.DATE)
FROM #T A
WHERE A.ID=T.ID
GROUP BY A.ID )
GROUP BY T.ID
select id, max(date) from NameOfYourTable group by id;

select the latest result based on DateTime field

I have a simple table with only 4 fields.
http://sqlfiddle.com/#!3/06d7d/1
CREATE TABLE Assessment (
id INTEGER IDENTITY(1,1) PRIMARY KEY,
personId INTEGER NOT NULL,
dateTaken DATETIME,
outcomeLevel VARCHAR(2)
)
INSERT INTO Assessment (personId, dateTaken, outcomeLevel)
VALUES (1, '2014-04-01', 'L1')
INSERT INTO Assessment (personId, dateTaken, outcomeLevel)
VALUES (1, '2014-04-05', 'L2')
INSERT INTO Assessment (personId, dateTaken, outcomeLevel)
VALUES (2, '2014-04-03', 'E3')
INSERT INTO Assessment (personId, dateTaken, outcomeLevel)
VALUES (2, '2014-04-07', 'L1')
I am trying to select for each "personId" their latest assessment result based on the dateTaken.
So my desired output for the following data would be.
[personId, outcomeLevel]
[1, L2]
[2, L1]
Thanks,
Danny
Try this:
;with cte as
(select personId pid, max(dateTaken) maxdate
from assessment
group by personId)
select personId, outcomeLevel
from assessment a
inner join cte c on a.personId = c.pid
where c.maxdate = a.dateTaken
order by a.personId
;with Cte as (Select personId,outcomeLevel, C= ROW_NUMBER()
over(PARTITION By personId Order By dateTaken desc)
From #Assessment
)
Select * from cte where C=1
Sample here
SELECT asst.personId,
asst.outcomeLevel
FROM dbo.Assessment asst
WHERE asst.dateTaken=(SELECT MAX(ast.dateTaken)
FROM assessment ast
WHERE asst.personid=ast.personId)
ORDER BY asst.personId
Result will be like this
personId outcomeLevel
1 L2
2 L1
Here is a possible solution using common table expression:
WITH cte AS (
SELECT
ROW_NUMBER() OVER (PARTITION BY personId ORDER BY dateTaken DESC) AS rn
, personId
, outcomeLevel
FROM
[dbo].[Assessment]
)
SELECT
personId
, outcomeLevel
FROM
cte
WHERE
rn = 1
About CTEs
A common table expression (CTE) can be thought of as a temporary result set that is defined within the execution scope of a single SELECT, INSERT, UPDATE, DELETE, or CREATE VIEW statement. A CTE is similar to a derived table in that it is not stored as an object and lasts only for the duration of the query. Unlike a derived table, a CTE can be self-referencing and can be referenced multiple times in the same query. From MSDN: Using Common Table Expressions
try this:
SELECT a.personId, a.outcomeLevel
FROM Assessment a
INNER JOIN
(
SELECT max(dateTaken) as datetaken1, personId
FROM Assessment
GROUP BY personId ) b
ON a.dateTaken = b.datetaken1
demo: http://sqlfiddle.com/#!3/06d7d/9
Idea is to first derive a table with the max dates per person and then join that with the original table on the date field so you can get the outcome level for this maxed date...
This should work perfectly without cte :
SELECT [Table4].[personId], [Table4].[outcomeLevel]
FROM (
SELECT [Table1].[personId]
FROM [Assessment] AS [Table1]
GROUP BY [Table1].[personId]
) AS [Table2]
CROSS APPLY (
SELECT TOP (1) [Table3].[personId], [Table3].[outcomeLevel], [Table3].[dateTaken]
FROM [Assessment] AS [Table3]
WHERE [Table2].[personId] = [Table3].[personId]
ORDER BY [Table3].[dateTaken] DESC
) AS [Table4]
ORDER BY [Table4].[dateTaken] DESC