I have the following problem:
I have an account table and an entries for account table.
account_id
account_name
entry_id
account_idfk
entry_date
entry_amount
Now I want to query all entries for all accounts in a given period. Eg. I want all Entries for all accounts from October 2008 - October 2009. If there are no entries for this account at all, or there are only entries in other timeperiods for this account, I want the account returned as well.
My current query works, if there are no entries at all, or there are entries for this timeperiod for this account. However - it leaves out the Accounts which have only entries for other timeperiods.
SELECT * FROM Account a
LEFT JOIN Entries e ON e.account_idfk = a.account_id
WHERE e.entry_date BETWEEN '2009-08-13' AND '2009-08-13'
OR e.entry_date IS NULL
I know that the problem is in the where clause - I eliminate all Accounts for which only entries in other time periods exist.
But I have no idea how to restate the query to get the desired result...
Thanks,
Martin
Move that condition to the join:
SELECT
*
FROM
Account a
LEFT JOIN Entries e ON
e.account_idfk = a.account_id
AND e.entry_date BETWEEN '2009-08-13' AND '2009-08-13'
What you see here is the difference between a join and a where condition. The join will only join rows that meet that condition. However, with a left join, you still return all the rows in the left table. With the where clause, you're filtering rows after the join. In this case, you only want to join entries where the date is 8/13/09 (or 13/8/09, for those across the pond), but you want to return all accounts. Therefore, the condition needs to go into the join clause, and not the where.
This often gets confused with any outer join, because with an inner join, the result is the same no matter if the condition is in the join or where clause. However, this does not mean that they are equivalent, as demonstrated by you today!
Related
I have tried every answer that I have found to finding the last record, and I have failed in getting a successful result. I currently have a query that lists active trailers. I am needing it to only show a single row for each trailer entry, where that row is based on a date in a joined table.
I have tables
trailer, company, equipment_group, movement, stop
In order to connect trailer to stop (which is where the date is), i have to join it to equipment group, which joins to movement, which then joins to stop.
I have tried using MAX and GROUP BY, and PARTITION BY, both of which error out.
I have tried many solutions here, as well as these
https://thoughtbot.com/blog/ordering-within-a-sql-group-by-clause
https://www.geeksengine.com/article/get-single-record-from-duplicates.html
It seems that all of these solutions have the date in the same table as the thing that they want to group by, which I do not.
SELECT
trailer.*
company.name,
equipment_group.currentmovement_id,
equipment_group.company_id,
movement.dest_stop_id, stop.location_id,
stop.*
FROM trailer
LEFT OUTER JOIN company ON (company.id = trailer.company_id)
LEFT OUTER JOIN equipment_group ON (equipment_group.id =
trailer.currenteqpgrpid)
LEFT OUTER JOIN movement ON (movement.id =
equipment_group.currentmovement_id)
LEFT OUTER JOIN stop ON (stop.id = movement.dest_stop_id)
WHERE trailer.is_active = 'A'
Using MAX and GROUP BY gives error "invalid in the select list... not contained in...aggregate function"
Welllllll, I never did end up figuring that out, but if I joined movements on to equipment group by two conditions, all is well. Each extra record was created by each company id.... company id is in EVERY table.
I have two tables that I want to join. Table1 has sales order, but it doesn’t have the name of the sales person. It only has employee ID. I have table2, that has the names of employees, and employeeID is common between the two tables. Normally I would use an inner join to get the name of the sales person from table2. The problem is that on table2, there are multiple entries for each employee. If they changed manager, or changed roles within the company, or perhaps went on FMLA, it creates a new row. Therefore, when I join the tables, it creates duplicates because of the multiple entries in table2. A sale shows 3 or 4 times in my results.
Select
a.state_name
,order_number
,a.employeeID
,b.Sales_Rep_Name
,a.order_date
from
table1 as A
Inner join table2 as B
On a.employeeid = b.employeeID
where
b.monthperiod = 'November' <-- If I remove this one it adds duplicates
Is there a way to not get these duplicates? I tried distinct but didn’t work. Probably because the rows have at least one column different. I was able to eliminate the duplicates when I added a where clause asking for last month on table 2, but I am in a situation where I need all months, not just one. I have to manually change the month in order to get the full year.
Any help would be appreciated. Thanks
Use a subquery to get list of distinct employee records and then query the sales table
I have two tables: Company and Contact, with a relationship of one-to-many.
I have another table Track which identifies some of the companies as parent companies to other companies.
I want to write a SQL query that selects the parent companies from Track and the amount of contacts that each parent has.
SELECT Track.ParentId, Count(Contact.companyId)
FROM Track
INNER JOIN Contact
ON Track.ParentId = Contact.companyId
GROUP BY Track.ParentId
however The result holds less records than when I run the following query:
SELECT DISTINCT Track.ParentId
FROM Track
I tried the first query with an added DISTINCT and it returned the same results (less then what it was meant to).
You're performing an INNER JOIN with the Contact table, which means that any rows from the first table (Track in this case) with no matches to the JOINed table will not show up in your results. Try using a LEFT OUTER JOIN instead.
The COUNT with Contact.companyId will only count rows where there is a match (Contact.companyId is not NULL). Since you're counting contacts that's fine as they will count as 0. If you were trying to count some other set of data and tried to do a COUNT on a specific column (rather than COUNT(*)) then any NULL values in that column would not count towards your total, which might or might not be what you want.
I used an INNER JOIN which returns only records that are identical in both tables.
To return all records from Track table, and records that match in the Contact table, I need to use LEFT JOIN.
I'm having a little confusion with a certain example.
I am supposed to list all orders and their corresponding details.
This is what I'm doing:
SELECT *
FROM Orders
LEFT OUTER JOIN [Order Details]
ON Orders.OrderID=[Order Details].OrderID;
This gives the number of rows = 2155.
Now the problem is, the number of rows in Orders Table is 830...how can left outer join create more rows ?
By definition of left outer join, all the rows from the left table are taken and matching records from the second table are added.
I checked the number of rows in the Order Details table..that is 2155.
Why is left outer join using all rows from Order Details table?
LEFT JOIN takes all the details from the table you define on the left side of the join and match records from the right table.
If there's no match, all columns of the right table have NULL values.
If there's a match, all matching records from the right table are returned. If your relationship is 1-to-many (as in your case), it means that there may be more than one record returned from the right table for each record in the left table.
LEFT OUTER JOIN will match all records in the right table, just as an INNER JOIN will. The difference is a LEFT JOIN will preserve records from the left table with no match in the right table.
In this scenario, all records in [ORDER DETAILS] have a corresponding entry in ORDERS, which is why the total number of records matches the number of rows in ORDER DETAILS
Based on the table descriptions, this is exactly what you want. Having an ORDER DETAIL without an ORDER would be a much more serious issue.
I am trying to do an OUTER JOIN, with multiple join conditions. Here is my query (I will explain issue below):
SELECT ad.*, cp.P_A, cp.P_B, cp.P_C
INTO #AggData3
FROM #AggData2 ad
LEFT OUTER JOIN #CompPriceTemp cp
ON ad.PART=cp.Part_No
and ad.[Month]=cp.[Month]
and ad.[Year]=cp.[Year]
GO
For each record in #AggData2, which is average price and volume by month for each part, I want to join the prices of the three competitors (A, B & C). Thus, I want to join based on Part, Month, and Year. Because some competitors don't offer all parts, I am using a LEFT OUTER JOIN. So, the resulting table (#AggData3), should have the exact same number of rows as the initial table (#AggData2), just with the three additional columns with competitor prices.
However, the new table (#AggData3), has ~35,000 more rows than #AggData2.
Any ideas why that is happening, and how to fix my query.
Because there are multiple rows in Table #CompPriceTemp that match to one row in #AggData2.
Is there one for each of three competitors perhaps? If that is so, then you need three joins, each to the same table, one for each of the 3 competitors?
But if there is supposed to be one row in #CompPriceTemp for each Month, Year, and product, with three separate columns one column for each competitor, then you have some bad data in there.
Wild guess:
ON ad.PART=cp.Part_No
and ad.[Month]=cp.[Month]
and ad.[Year]=cp.[Year]
This query does not uniquely identify rows in CP. Or CP has ~35000 duplicate rows.
Are you sure that you have only one matching row in CompPriceTemp for every single row in AggData2 ?