Select newest data from id A - sql

I have the data like this:
ID TRANS_ID CREATED_DATE STATUS
----------------------------------------------
1 AA 2017-05-19 02:00:00 WAITING
2 AA 2017-05-20 02:00:00 IN_PROCESS
3 BB 2017-05-19 02:00:00 WAITING
4 CC 2017-05-19 02:00:00 WAITING
5 CC 2017-05-20 02:00:00 IN_PROCESS
I would like to show the data in table view like this one :
ID TRANS_ID CREATED_DATE STATUS
----------------------------------------------
2 AA 2017-05-20 02:00:00 IN_PROCESS
3 BB 2017-05-19 02:00:00 WAITING
5 CC 2017-05-20 02:00:00 IN_PROCESS
I take the newest data from each trans_id and I run this query but it does not work
select id, max(created_date), trans_id, status
from table_a
group by a.transaction_id

One option is to filter with a correlated subquery:
select t.*
from mytable t
where t.created_date = (
select max(t1.created_date) from mytable t1 where t1.trans_id = t.trans_id
)
Alternatively, you can use window functions:
select id, trans_id, created_date, status
from (
select t.*, rank() over(partition by trans_id order by created_date desc) rn
from mytable t
) t
where rn = 1
This allows top ties, if any. If you want no ties, you can use row_number() instead of rank().

Related

Get max date for each from either of 2 columns

I have a table like below
AID BID CDate
-----------------------------------------------------
1 2 2018-11-01 00:00:00.000
8 1 2018-11-08 00:00:00.000
1 3 2018-11-09 00:00:00.000
7 1 2018-11-15 00:00:00.000
6 1 2018-12-24 00:00:00.000
2 5 2018-11-02 00:00:00.000
2 7 2018-12-15 00:00:00.000
And I am trying to get a result set as follows
ID MaxDate
-------------------
1 2018-12-24 00:00:00.000
2 2018-12-15 00:00:00.000
Each value in the id columns(AID,BID) should return the max of CDate .
ex: in the case of 1, its max CDate is 2018-12-24 00:00:00.000 (here 1 appears under BID)
in the case of 2 , max date is 2018-12-15 00:00:00.000 . (here 2 is under AID)
I tried the following.
1.
select
g.AID,g.BID,
max(g.CDate) as 'LastDate'
from dbo.TT g
inner join
(select AID,BID,max(CDate) as maxdate
from dbo.TT
group by AID,BID)a
on (a.AID=g.AID or a.BID=g.BID)
and a.maxdate=g.CDate
group by g.AID,g.BID
and 2.
SELECT
AID,
CDate
FROM (
SELECT
*,
max_date = MAX(CDate) OVER (PARTITION BY [AID])
FROM dbo.TT
) AS s
WHERE CDate= max_date
Please suggest a 3rd solution.
You can assemble the data in a table expression first, and the compute the max for each value is simple. For example:
select
id, max(cdate)
from (
select aid as id, cdate from t
union all
select bid, cdate from t
) x
group by id
You seem to only care about values that are in both columns. If this interpretation is correct, then:
select id, max(cdate)
from ((select aid as id, cdate, 1 as is_a, 0 as is_b
from t
) union all
(select bid as id, cdate, 1 as is_a, 0 as is_b
from t
)
) ab
group by id
having max(is_a) = 1 and max(is_b) = 1;

Query and Partition By clause group by window

I've the following code
declare #test table (id int, [Status] int, [Date] date)
insert into #test (Id,[Status],[Date]) VALUES
(1,1,'2018-01-01'),
(2,1,'2018-01-01'),
(1,1,'2017-11-01'),
(1,2,'2017-10-01'),
(1,1,'2017-09-01'),
(2,2,'2017-01-01'),
(1,1,'2017-08-01'),
(1,1,'2017-07-01'),
(1,1,'2017-06-01'),
(1,2,'2017-05-01'),
(1,1,'2017-04-01'),
(1,1,'2017-03-01'),
(1,1,'2017-01-01')
SELECT
id,
[Status],
MIN([Date]) OVER (PARTITION BY id,[Status] ORDER BY [Date],id,[Status] ) as WindowStart,
max([Date]) OVER (PARTITION BY id,[Status] ORDER BY [Date],id,[Status]) as WindowEnd,
COUNT(*) OVER (PARTITION BY id,[Status] ORDER BY [Date],id,[Status] ) as total
from #test
But the result is this:
id Status WindowStart WindowEnd total
1 1 2017-01-01 2017-01-01 1
1 1 2017-01-01 2017-03-01 2
1 1 2017-01-01 2017-04-01 3
1 1 2017-01-01 2017-06-01 4
1 1 2017-01-01 2017-07-01 5
1 1 2017-01-01 2017-08-01 6
1 1 2017-01-01 2017-09-01 7
1 1 2017-01-01 2017-11-01 8
1 1 2017-01-01 2018-01-01 9
1 2 2017-05-01 2017-05-01 1
1 2 2017-05-01 2017-10-01 2
2 1 2018-01-01 2018-01-01 1
2 2 2017-01-01 2017-01-01 1
And I need to be grouped by window like this.
id Status WindowStart WindowEnd total
1 1 2017-01-01 2017-04-01 3
1 2 2017-05-01 2017-05-01 1
1 1 2017-06-01 2017-09-01 4
1 2 2017-10-01 2017-10-01 1
1 1 2017-11-01 2018-01-01 2
2 1 2018-01-01 2018-01-01 1
2 2 2017-01-01 2017-01-01 1
The first group for the id= 1 Status = 1 should end at the first row with Status = 2 (2017-05-01) so the total is 3 and then start again from the 2017-06-01 to 2017-09-01 with a total of 4 rows.
How can get this done?
This is a "classic" Groups and Island issue. There's probably 1000's of answers for these on the Internet.
This works for what you're after, however, try having a bit more of a research before hand. :)
WITH Groups AS(
SELECT t.*,
ROW_NUMBER() OVER (PARTITION BY id ORDER BY [Date]) -
ROW_NUMBER() OVER (PARTITION BY id, [status] ORDER BY [Date]) AS Grp
FROM #test t)
SELECT G.id,
G.[Status],
MIN([Date]) AS WindowStart,
MAX([date]) AS WindowsEnd,
COUNT(*) AS Total
FROM Groups G
GROUP BY G.id,
G.[Status],
G.Grp
ORDER BY G.id, WindowStart;
Note, that the ordering of your last 2 lines is the other way round in this solution; it seems you're ordering ASCENDING for id 1, for DESCENDING for id 2 in your expected results.
Here is one way using LAG function
;WITH cte
AS (SELECT *,
grp = Sum(CASE WHEN prev_val = Status THEN 0 ELSE 1 END)
OVER(partition BY id ORDER BY Date)
FROM (SELECT *,
prev_val = Lag(Status)OVER(partition BY id ORDER BY Date)
FROM #test) a)
SELECT id,
Status,
WindowStart = Min(date),
WindowEnd = Max(date),
Total = Count(*)
FROM cte
GROUP BY id, Status, grp
Using lag function first find the previous status of each date, then using Sum over() create a group by incrementing the number only when there is a change in status.

select latest record for each battery using SQL with count

BatteryId TimeStamp Temprature
1 2017-02-13 12:16:14.000 23
1 2016-02-13 12:13:14.000 21
1 2015-01-13 12:16:14.000 19
2 2017-02-11 12:16:14.000 22
2 2016-02-13 12:16:14.000 16
3 2017-02-13 11:16:14.000 12
3 2016-02-13 12:15:14.000 25
I have table with multiple records for each battery as above
following sql query is returning latest record for each battery
SELECT * FROM (SELECT BatteryId, Timestamp, Temperature
ROW_NUMBER() OVER(PARTITION BY BatteryId ORDER BY timestamp DESC)
AS N FROM tblBattery) AS TT WHERE N = 1
as
BatteryId TimeStamp Temprature
1 2017-02-13 12:16:14.000 23
2 2017-02-11 12:16:14.000 22
3 2017-02-13 11:16:14.000 12
How I can add Count for each BatteryId, Here is what I need
BatteryId TimeStamp Temprature Count
1 2017-02-13 12:16:14.000 23 3
2 2017-02-11 12:16:14.000 22 2
3 2017-02-13 11:16:14.000 12 2
Use the count window function.
SELECT * FROM
(SELECT BatteryId, Timestamp, Temperature,
ROW_NUMBER() OVER(PARTITION BY BatteryId ORDER BY timestamp DESC) AS N,
COUNT(*) OVER(PARTITION BY BatteryId) as Cnt
FROM tblBattery) TT
WHERE N = 1
Hoping, i understood your problem correctly.
Please check if below query can help you.
SELECT *
FROM
(SELECT BatteryId,
TIMESTAMP,
Temperature , ROW_NUMBER() OVER(PARTITION BY BatteryId ORDER BY TIMESTAMP DESC) AS N ,
COUNT(0) OVER(PARTITION BY BatteryId ) CNT
FROM tblBattery
) AS TT
WHERE N = 1;
Add a sub query before you perform the PARTITION BY
SELECT *
FROM (SELECT
BatteryId
,Timestamp
,Temperature
,Count
,ROW_NUMBER() OVER(PARTITION BY BatteryId ORDER BY timestamp DESC) AS N
FROM (SELECT *, COUNT(BatteryId) As Count FROM tblBattery GROUP BY BatteryId)) AS TT WHERE N = 1
This should solve your issue.

Rank on each group without using derived table

AccountNo Account Name Transaction time Transaction Amount
A01 Name 1 01-01-2016 04:00:00 5000
A01 Name 1 01-01-2016 07:30:00 3500
A01 Name 1 01-01-2016 09:30:00 3500
A01 Name 1 01-01-2016 12:00:00 6500
A01 Name 1 02-02-2016 06:00:00 4000
A01 Name 1 01-02-2016 08:30:00 8000
A01 Name 1 01-02-2016 09:30:00 8000
A02 Name 2 05-01-2016 04:00:00 2000
A02 Name 2 05-01-2016 07:30:00 8500
A02 Name 2 08-02-2016 06:00:00 1000
A02 Name 2 09-02-2016 08:30:00 9000
I need a query without using derived table to get the records which has latest transactions for each account. Query should support Oracle or TERADATA.
Using Derived table I am able to solve it as below:
SELECT a.accountno,
a.account_name,
a.transaction_time,
a.transaction_amount
FROM acct_details a,
(SELECT accountno,
Max(transaction_time) AS Transaction_time
FROM acct_details
GROUP BY accountno) b
WHERE a.accountno = b.accountno
AND a.transaction_time = b.transaction_time;
Thanks for Help!!
Why without Derived Table?
In Teradata there's QUALIFY, but it's no Standard SQL:
SELECT *
FROM acct_details
QUALIFY RANK() OVER (PARTITION BY accountno
ORDER BY transaction_time DESC) = 1;
In Oracle you need a Derived Table:
SELECT a.accountno,
a.account_name,
a.transaction_time,
a.transaction_amount,
FROM
(
SELECT a.accountno,
a.account_name,
a.transaction_time,
a.transaction_amount,
RANK() OVER (PARTITION BY accountno
ORDER BY transaction_time DESC) as rnk
FROM acct_details a
) a
WHERE rnk = 1
You might also try Oracle's MAX/KEEP:
SELECT a.accountno,
MAX (a.account_name) KEEP (DENSE_RANK FIRST ORDER BY a.transaction_time desc) ,
MAX(a.transaction_time),
MAX (a.transaction_amount) KEEP (DENSE_RANK FIRST ORDER BY a.transaction_time desc) FROM acct_details a
GROUP BY a.accountno
You can do it using an inline view like so:
SELECT a.accountno,
a.account_name,
a.transaction_time,
a.transaction_amount
FROM (SELECT accountno,
account_name,
transaction_time,
transaction_amount,
Max(transaction_time) OVER (PARTITION BY accountno) AS max_transaction_time
FROM acct_details) a
WHERE a.transaction_time = a.max_transaction_time;
This will work on Oracle and I think it should also work on Teradata.
I assume here that when you say you want to avoid a derived table, you mean you want to avoid joining the table to itself.
try this query:
SELECT [Transaction time],[Account Name],
lead([Transaction time ],1,1) OVER (ORDER BY [Account Name])
FROM [dbo].[acct_details]
this should help you

SQL First Time In Last Time Out (Transact-SQL..preferrably)

I have the following table which contains all the time in and time out of people:
CREATE TABLE test (
timecardid INT
, trandate DATE
, employeeid INT
, trantime TIME
, Trantype VARCHAR(1)
, Projcode VARCHAR(3)
)
The task is to get all the earliest trantime with trantype A (perhaps using MIN) and the latest trantime with trantype Z (Using Max), all of which in that trandate (ie. trantype A for july 17 is 8:00 AM and trantype Z for july 17 is 7:00PM).
the problem is, the output should be in the same format as the table where it's coming from, meaning that I have to leave this data and filter out the rest (that aren't the earliest and latest in/out for that date, per employee)
My current solution is to use two different select commands to get all earliest, then get all the latest. then combine them both.
I was wondering though, is there a much simpler, single string solution?
Thank you very much.
EDIT (I apologize, here is the sample. Server is SQL Server 2008):
Timecardid | Trandate | employeeid | trantime | trantype | Projcode
1 2013-04-01 1 8:00:00 A SAMPLE1
2 2013-04-01 1 9:00:00 A SAMPLE1
3 2013-04-01 2 7:00:00 A SAMPLE1
4 2013-04-01 2 6:59:59 A SAMPLE1
5 2013-04-01 1 17:00:00 Z SAMPLE1
6 2013-04-01 1 17:19:00 Z SAMPLE1
7 2013-04-01 2 17:00:00 Z SAMPLE1
8 2013-04-02 1 8:00:00 A SAMPLE1
9 2013-04-02 1 9:00:00 A SAMPLE1
10 2013-04-02 2 7:00:58 A SAMPLE1
11 2013-04-02 2 18:00:00 Z SAMPLE1
12 2013-04-02 2 18:00:01 Z SAMPLE1
13 2013-04-02 1 20:00:00 Z SAMPLE1
Expected Results (the earliest in and the latest out per day, per employee, in a select command):
Timecardid | Trandate | employeeid | trantime | trantype | Projcode
1 2013-04-01 1 8:00:00 A SAMPLE1
4 2013-04-01 2 6:59:59 A SAMPLE1
6 2013-04-01 1 17:19:00 Z SAMPLE1
7 2013-04-01 2 17:00:00 Z SAMPLE1
8 2013-04-02 1 8:00:00 A SAMPLE1
10 2013-04-02 2 7:00:58 A SAMPLE1
12 2013-04-02 2 18:00:01 Z SAMPLE1
13 2013-04-02 1 20:00:00 Z SAMPLE1
Thank you very much
Perhaps this is what you're looking for:
select
t.*
from
test t
where
trantime in (
(select min(trantime) from test t1 where t1.trandate = t.trandate and trantype = 'A'),
(select max(trantime) from test t2 where t2.trandate = t.trandate and trantype = 'Z')
)
Changing my answer to account for the "per employee" requirement:
;WITH EarliestIn AS
(
SELECT trandate, employeeid, min(trantime) AS EarliestTimeIn
FROM test
WHERE trantype = 'A'
GROUP BY trandate, employeeid
),
LatestOut AS
(
SELECT trandate, employeeid, max(trantime) AS LatestTimeOut
FROM test
WHERE trantype = 'Z'
GROUP BY trandate, employeeid
)
SELECT *
FROM test t
WHERE
EXISTS (SELECT * FROM EarliestIn WHERE t.trandate = EarliestIn.trandate AND t.employeeid = EarliestIn.employeeid AND t.trantime = EarliestIn.EarliestTimeIn)
OR EXISTS (SELECT * FROM LatestOut WHERE t.trandate = LatestOut.trandate AND t.employeeid = LatestOut.employeeid AND t.trantime = LatestOut.LatestTimeOut)
Assuming timecardid column is PK or unique, and if I understand it correctly, I would do something like
DECLARE #date DATE
SET #date = '2013-07-01'
SELECT
T0.*
FROM
(SELECT DISTINCT employeeid FROM test) E
CROSS APPLY (
SELECT TOP 1
T.timecardid
FROM
test T
WHERE
T.trandate = #date
AND T.Trantype = 'A'
AND T.employeeid = E.employeeid
ORDER BY T.trantime
UNION ALL
SELECT TOP 1
T.timecardid
FROM
test T
WHERE
T.trandate = #date
AND T.Trantype = 'Z'
AND T.employeeid = E.employeeid
ORDER BY T.trantime DESC
) V
JOIN test T0 ON T0.timecardid = V.timecardid
Appropriate indexes should be set for the table, if you aware of performance.
If you're using SQL server 2012, you can use LAG/LEAD to find the max and min rows in a fairly concise way;
WITH cte AS (
SELECT *,
LAG(timecardid) OVER (PARTITION BY trandate,employeeid,trantype ORDER BY trantime) lagid,
LEAD(timecardid) OVER (PARTITION BY trandate,employeeid,trantype ORDER BY trantime) leadid
FROM test
)
SELECT timecardid,trandate,employeeid,trantime,trantype,projcode
FROM cte
WHERE trantype='A' AND lagid IS NULL
OR trantype='Z' AND leadid IS NULL;
An SQLfiddle to test with.
I would use ROW_NUMBER to sort out the rows you want to select:
;with Ordered as (
select *,
ROW_NUMBER() OVER (PARTITION BY Trandate,employeeid,trantype
ORDER BY trantime ASC) as rnEarly,
ROW_NUMBER() OVER (PARTITION BY Trandate,employeeid,trantype
ORDER BY trantime DESC) as rnLate
from
Test
)
select * from Ordered
where
(rnEarly = 1 and trantype='A') or
(rnLate = 1 and trantype='Z')
order by TimecardId
(SQLFiddle)
It produces the results you've requested, and I think it's quite readable. The reason that trantype is included in the PARTITION BY clauses is so that A and Z values receive separate numbering.