I have an MS Access database with the two tables, Asset and Transaction. The schema looks like this:
Table ASSET
Key Date1 AType FieldB FieldC ...
A 2023.01.01 T1
B 2022.01.01 T1
C 2023.01.01 T2
.
.
TABLE TRANSACTION
Date2 Key TType1 TType2 TType3 FieldOfInterest ...
2022.05.31 A 1 1 1 10
2022.08.31 A 1 1 1 40
2022.08.31 A 1 2 1 41
2022.09.31 A 1 1 1 30
2022.07.31 A 1 1 1 30
2022.06.31 A 1 1 1 20
2022.10.31 A 1 1 1 45
2022.12.31 A 2 1 1 50
2022.11.31 A 1 2 1 47
2022.05.23 B 2 1 1 30
2022.05.01 B 1 1 1 10
2022.05.12 B 1 2 1 20
.
.
.
The ASSET table has a PK (Key).
The TRANSACTION table has a composite key that is (Key, Date2, Type1, Type2, Type3).
Given the above tables let's see an example:
Input1 = 2022.04.01
Input2 = 2022.08.31
Desired result:
Key FieldOfInterest
A 41
because if the Transactions in scope was to be ordered by Date2, TType1, TType2, TType3 all ascending then the record having FieldOfInterest = 41 would be the last one.
Note that Asset B is not in scope due to Asset.Date1 < Input1, neither is Asset C because AType != T1. Ultimately I am curious about the SUM(FieldOfInterest) of all the last transactions belonging to an Asset that is in scope determined by the input variables.
The following query has so far provided the right results but after upgrading to a newer MS Access version, the LAST() operation is no longer reliably returning the row which is the latest addition to the Transaction table.
I have several input values but the most important ones are two dates, lets call them InputDate1 and
InputDate2.
This is how it worked so far:
SELECT Asset.AType, Last(FieldOfInterest) AS CurrentValue ,Asset.Key
FROM Transaction
INNER JOIN Asset ON Transaction.Key = Asset.Key
WHERE Transaction.Date2 <= InputDate2 And Asset.Date1 >= InputDate1
GROUP BY Asset.Key, Asset.AType
HAVING Asset.AType='T1'
It is known that the grouped records are not guaranteed to be in any order. Obviously it is a mistake to rely on the order of the records of the group by operation will always keep the original table order but lets just ignore this for now.
I have been struggling to come up with the right way to do the following:
join the Asset and Transaction tables on Asset.Key = Transaction.Key
filter by Asset.Date1 >= InputDate1 AND Transaction.Date2 <= InputDate2
then I need to select one record for all Transaction.Key where Date2 and TType1 and TType2 and TType3 has the highest value. (this represents the actual last record for given Key)
As far as I know there is no way to order records within a group by clause which is unfortunate.
I have tried Ranking, but the Transactions table is large (800k rows) and the performance was very slow, I need something faster than this. The following are an example of three saved queries that I wrote and chained together but the performance is very disappointing probably due to the ranking step.
-- Saved query step_1
SELECT Asset.*, Transaction.*
FROM Transaction
INNER JOIN Asset ON Transaction.Key = Asset.Key
WHERE Transaction.Date2 <= 44926
AND Asset.Date1 >= 44562
AND Asset.aType = 'T1'
-- Saved query step_2
SELECT tr.FieldOfInterest, (SELECT Count(*) FROM
(SELECT tr2.Transaction.Key, tr2.Date2, tr2.Transaction.tType1, tr2.tType2, tr2.tType3 FROM step_1 AS tr2) AS tr1
WHERE (tr1.Date2 > tr.Date2 OR
(tr1.Date2 = tr.Date2 AND tr1.tType1 > tr.Transaction.tType1) OR
(tr1.Date2 = tr.Date2 AND tr1.tType1 = tr.Transaction.tType1 AND tr1.tType2 > tr.tType2) OR
(tr1.Date2 = tr.Date2 AND tr1.tType1 = tr.Transaction.tType1 AND tr1.tType2 = tr.tType2 AND tr1.tType3 > tr.tType3))
AND tr1.Key = tr.Transaction.Key)+1 AS Rank
FROM step_1 AS tr
-- Saved query step_3
SELECT SUM(FieldOfInterest) FROM step_2
WHERE Rank = 1
I hope I am being clear enough so that I can get some useful recommendations. I've been stuck with this for weeks now and really don't know what to do about it. I am open for any suggestions.
Reading the following specification
then I need to select one record for all Transaction.Key where Date2 and TType1 and TType2 and TType3 has the highest value. (this represents the actual last record for given Key)
Consider a simple aggregation for step 2 to retrieve the max values then in step 3 join all fields to first query.
Step 1 (rewritten to avoid name collision and too many columns)
SELECT a.[Key] AS Asset_Key, a.Date1, a.AType,
t.[Key] AS Transaction_Key, t.Date2,
t.TType1, t.TType2, t.TType3, t.FieldOfInterest
FROM Transaction t
INNER JOIN Asset a ON a.[Key] = a.[Key]
WHERE t.Date2 <= 44926
AND a.Date1 >= 44562
AND a.AType = 'T1'
Step 2
SELECT Transaction_Key,
MAX(Date2) AS Max_Date2,
MAX(TType1) AS TType1,
MAX(TType2) AS TType2,
MAX(TType3) AS TType3
FROM step_1
GROUP Transaction_Key
Step 3
SELECT s1.*
FROM step_1 s1
INNER JOIN step_2 s2
ON s1.Transaction_Key = s2.Transaction_Key
AND s1.Date2 = s2.Max_Date2
AND s1.TType1 = s2.Max_TType1
AND s1.TType2 = s2.Max_TType2
AND s1.TType3 = s2.Max_TType3
Related
I want to compare data of a table with its other records. The count of rows with a specific condition has to match the count of rows without the where clause but on the same grouping.
Below is the table
-------------
id name time status
1 John 10 C
2 Alex 10 R
3 Dan 10 C
4 Tim 11 C
5 Tom 11 C
Output should be time = 11 as the count for grouping on time column is different when a where clause is added on status = 'C'
SELECT q1.time
FROM (SELECT time,
Count(id)
FROM table
GROUP BY time) AS q1
INNER JOIN (SELECT time,
Count(id)
FROM table
WHERE status = 'C'
GROUP BY time) AS q2
ON q1.time = q2.time
WHERE q1.count = q2.count
This is giving the desired output but is there a better and efficient way to get the desired result?
Are you looking for this :
select t.*
from table t
where not exists (select 1 from table t1 where t1.time = t.time and t1.status <> 'C');
However you can do :
select time
from table t
group by time
having sum (case when status <> 'c' then 1 else 0 end ) = 0;
If you want the times where the rows all satisfy the where clause, then in Postgres, you can express this as:
select time
from t
group by time
having count(*) = count(*) filter (where status = 'C');
I tried to solve one problem but without success.
I have two list of number
{1,2,3,4}
{5,6,7,8,9}
And I have table
ID Number
1 1
1 2
1 7
1 2
1 6
2 8
2 7
2 3
2 9
Now I need to count how many times number from second list come after number from first list but I should count only one by one id
in example table above result should be 2
three matched pars but because we have only two different IDs result is 2 instead 3
Pars:
1 2
1 7
1 2
1 6
2 3
2 9
note. I work with MSSQL
Edit. There is one more column Date which determined order
Edit2 - Solution
i write this query
SELECT * FROM table t
left JOIN table tt ON tt.ID = t.ID
AND tt.Date > t.Date
AND t.Number IN (1,2,3,4)
AND tt.Number IN (6,7,8,9)
And after this I had a plan to group by id and use only one match for each id but execution take a lot time
Here is a query that would do it:
select a.id, min(a.number) as a, min(b.number) as b
from mytable a
inner join mytable b
on a.id = b.id
and a.date < b.date
and b.number in (5,6,7,8,9)
where a.number in (1,2,3,4)
group by a.id
Output is:
id a b
1 1 6
2 3 9
So the two pairs are output each on one line, with the value a belonging to the first group of numbers, and the value of column b to the second group.
Here is a fiddle
Comments on attempt (edit 2 to question)
Later you added a query attempt to your question. Some comments about that attempt:
You don't need a left join because you really want to have a match for both values. inner join has in general better performance, so use that.
The condition t.Number IN (1,2,3,4) does not belong in the on clause. In combination with a left join the result will include t records that violate this condition. It should be put in the where clause.
Your concern about performance may be warranted, but can be resolved by adding a useful index on your table, i.e. on (id, number, date) or (id, date, number)
I want to identify the users who visited section a and then subsequently visited b. Given the following data structure. The table contains 300,000 rows and updates daily with approx. 8,000 rows:
**USERID** **VISITID** **SECTION** Desired Solution--> **Conversion**
1 1 a 0
1 2 a 0
2 1 b 0
2 1 b 0
2 1 b 0
1 3 b 1
Ideally I want a new column that flags the visit to section b. For example on the third visit User 1 visited section b for the first time. I was attempting to do this using a CASE WHEN statement but after many failed attempts I am not sure it is even possible with CASE WHEN and feel that I should take a different approach, I am just not sure what that approach should be. I do also have a date column at my disposal.
Any suggestions on a new way to approach the problem would be appreciated. Thanks!
Correlated sub-queries should be avoided at all cost when working with Redshift. Keep in mind there are no indexes for Redshift so you'd have to rescan and restitch the column data back together for each value in the parent resulting in an O(n^2) operation (in this particular case going from 300 thousand values scanned to 90 billion).
The best approach when you are looking to span a series of rows is to use an analytic function. There are a couple of options depending on how your data is structured but in the simplest case, you could use something like
select case
when section != lag(section) over (partition by userid order by visitid)
then 1
else 0
end
from ...
This assumes that your data for userid 2 increments the visitid as below. If not, you could also order by your timestamp column
**USERID** **VISITID** **SECTION** Desired Solution--> **Conversion**
1 1 a 0
1 2 a 0
2 1 b 0
2 *2* b 0
2 *3* b 0
1 3 b 1
select t.*, case when v.ts is null then 0 else 1 end as conversion
from tbl t
left join (select *
from tbl x
where section = 'b'
and exists (select 1
from tbl y
where y.userid = x.userid
and y.section = 'a'
and y.ts < x.ts)) v
on t.userid = v.userid
and t.visitid = v.visitid
and t.section = v.section
Fiddle:
http://sqlfiddle.com/#!15/5b954/5/0
I added sample timestamp data as that field is necessary to determine whether a comes before b or after b.
To incorporate analytic functions you could use:
(I've also made it so that only the first occurrence of B (after an A) will get flagged with the 1)
select t.*,
case
when v.first_b_after_a is not null
then 1
else 0
end as conversion
from tbl t
left join (select userid, min(ts) as first_b_after_a
from (select t.*,
sum( case when t.section = 'a' then 1 end)
over( partition by userid
order by ts ) as a_sum
from tbl t) x
where section = 'b'
and a_sum is not null
group by userid) v
on t.userid = v.userid
and t.ts = v.first_b_after_a
Fiddle: http://sqlfiddle.com/#!1/fa88f/2/0
After many attempts I have failed at this and hoping someone can help. The query returns every entry a user makes when items are made in the factory against and order number. For example
Order Number Entry type Quantity
3000 1 1000
3000 1 500
3000 2 300
3000 2 100
4000 2 1000
5000 1 1000
What I want to the query do is to return filter the results like this
If the order number has an entry type 1 and 2 return the row which is type 1 only
otherwise just return row whatever the type is for that order number.
So the above would end up:
Order Number Entry type Quantity
3000 1 1000
3000 1 500
4000 2 1000
5000 1 1000
Currently my query (DB2, in very basic terms looks like this ) and was correct until a change request came through!
Select * from bookings where type=1 or type=2
thanks!
select * from bookings
left outer join (
select order_number,
max(case when type=1 then 1 else 0 end) +
max(case when type=2 then 1 else 0 end) as type_1_and_2
from bookings
group by order_number
) has_1_and_2 on
type_1_and_2 = 2
has_1_and_2.order_number = bookings.order_number
where
bookings.type = 1 or
has_1_and_2.order_number is null
Find all the orders that have both type 1 and type 2, and then join it.
If the row matched the join, only return it if it is type 1
If the row did not match the join (has_type_2.order_number is null) return it no matter what the type is.
A "common table expression" [CTE] can often simplify your logic. You can think of it as a way to break a complex problem into conceptual steps. In the example below, you can think of g as the name of the result set of the CTE, which will then be joined to
WITH g as
( SELECT order_number, min(type) as low_type
FROM bookings
GROUP BY order_number
)
SELECT b.*
FROM g
JOIN bookings b ON g.order_number = b.order_number
AND g.low_type = b.type
The JOIN ON conditions will work so that if both types are present then low_type will be 1, and only that type of record will be chosen. If there is only one type it will be identical to low_type.
This should work fine as long as 1 and 2 are the only types allowed in the bookings table. If not then you can simply add a WHERE clause in the CTE and in the outer SELECT.
Let's say I have this table:
|Fld | Number|
1 5
2 2
And I want to make a select that retrieves as many Fld as the Number field has:
|Fld |
1
1
1
1
1
2
2
How can I achieve this? I was thinking about making a temporary table and instert data based on the Number, but I was wondering if this could be done with a single Select statement.
PS: I'm new to SQL
You can join with a numbers table:
SELECT Fld
FROM yourtable
JOIN Numbers
ON yourtable.Number <= Numbers.Number
A numbers table is just a table with a list of numbers:
Number
1
2
3
etc...
Not an great solution (since you still query your table twice, but maybe you can work from it)
SELECT t1.fld, t1.number
FROM table t1, (
SELECT ROWNUM number FROM dual
CONNECT BY LEVEL <= (SELECT MAX(number) FROM t1)) t2
WHERE t2.number<=t1.number
It generates maximum amount of rows needed and then filters it by each row.
I don't know if your RDBMS version supports it (although I rather suspect it does), but here is a recursive version:
WITH remaining (fld, times) as (SELECT fld, 1
FROM <table>
UNION ALL
SELECT a.fld, a.times + 1
FROM remaining as a
JOIN <table> as b
ON b.fld = a.fld
AND b.number > a.times)
SELECT fld
FROM remaining
ORDER BY fld
Given your source data table, it outputs this (count included for verification):
fld times
=============
1 1
1 2
1 3
1 4
1 5
2 1
2 2