I have a table here in which I want to write a SELECT query in SQL Server that allows me to get the following:
For each unique combination of SalesPerson x Country, get only the rows with the latest Upload_DateTime
However, I am trying to do a group-by and inner join, but to no avail. My code is something like this:
SELECT t1.[SalesPerson], t1.[Country], MAX(t1.[Upload_DateTime]) as [Upload_DateTime]
FROM [dbo].[CommentTable] AS t1
GROUP BY t1.[SalesPerson], t1.[Country]
INNER JOIN SELECT * FROM [dbo].[CommentTable] as t2 ON t1.[SalesPerson] = t2.[SalesPerson], t1.[Country] = t2.[Country]
It seems like the GROUP BY needs to be done outside of the INNER JOIN? How does that work? I get an error when I run the query and it seems my SQL is not right.
Basically, this subquery will fetch the person, the country and the latest date:
SELECT
SalesPerson, Country, MAX(uplodaed_datetime)
FROM CommentTable
GROUP BY SalesPerson, Country;
This can be used on a lot of ways (for example with JOIN or with an IN clause).
The main query will add the remaing columns to the result.
Since you tried a JOIN, here the JOIN option:
SELECT
c.id, c.SalesPerson, c.Country,
c.Comment, c.uplodaed_datetime
FROM
CommentTable AS c
INNER JOIN
(SELECT
SalesPerson, Country,
MAX(uplodaed_datetime) AS uplodaed_datetime
FROM CommentTable
GROUP BY SalesPerson, Country) AS sub
ON c.SalesPerson = sub.SalesPerson
AND c.Country = sub.Country
AND c.uplodaed_datetime = sub.uplodaed_datetime
ORDER BY c.id;
Try out: db<>fiddle
Related
I am trying to select which co-ordinates from OA table are NOT found in the CUSTOMER address table.
SELECT DISTINCT
OA.CO_ORDS
FROM
CUSTOMER
INNER JOIN
OA ON customer.address=oa.co_ords
ORDER BY ID ASC;
Returns the co-ordinates which ARE in the customer table. How do I return those that are not in the customer table?
Am I also able to COUNT how many of customers are is in each co-ordinate (The co-ords are not specific and not accurate, this is purely for query testing only)
SELECT DISTINCT
OA.CO_ORDS
FROM
CUSTOMER
INNER JOIN
OA ON customer.address=oa.co_ords
ORDER BY ID ASC;
We can use NOT EXISTS to find those co-ordinates which don't appear in the customer table:
SELECT co_ords
FROM oa
WHERE
NOT EXISTS
(SELECT 1 FROM customers
WHERE address = oa.co_ords)
ORDER BY id;
In order to count how many customers belong to a certain co-ordinate, we can use COUNT with GROUP BY, something like this:
SELECT c.address, COUNT(*)
FROM customers c
JOIN oa
ON c.address = oa.co_ords
GROUP BY c.address;
It could be better to count a specific column instead of *.
It could also be better to use an IN clause instead of JOIN the tables:
SELECT c.address, COUNT(*)
FROM customers c
WHERE address IN
(SELECT co_ords FROM oa)
GROUP BY c.address;
Such details depend on your exact table structure, you should please try this out or provide more details.
You could also do:
SELECT co_ords
FROM oa
MINUS
SELECT address
FROM customers;
which can sometimes be faster than doing an anti-join. Note that MINUS does a distinct on the resultset.
I am struggling with taking a Count() from one table and dividing it by a correlating number from a different table in Microsoft SQL Server.
Here is a fictional example of what I'm trying to do
Lets say I have a table of orders. One column in there is states.
I have a second table that has a column for states, and second column for each states population.
I'd like to find the order per population for each sate, but I have struggled to get my query right.
Here is what I have so far:
SELECT Orders.State, Count(*)/
(SELECT StatePopulations.Population FROM Orders INNER JOIN StatePopulations
on Orders.State = StatePopulations.State
WHERE Orders.state = StatePopulations.State )
FROM Orders INNER JOIN StatePopulations
ON Orders.state = StatePopulations.State
GROUP BY Orders.state
So far I'm contending with an error that says my sub query is returning multiple results for each state, but I'm newer to SQL and don't know how to overcome it.
If you really want a correlated sub-query, then this should do it...
(You don't need to join both table in either the inner or outer query, the correlation in the inner query's where clause does the 'join'.)
SELECT
Orders.state,
COUNT(*) / (SELECT population FROM StatePopulation WHERE state = Orders.state)
FROM
Orders
GROUP BY
Orders.state
Personally, I'd just join them and use MAX()...
SELECT
Orders.state,
COUNT(*) / MAX(StatePopulation.population)
FROM
Orders
INNER JOIN
StatePopulation
StatePopulation.state = Orders.state
GROUP BY
Orders.state
Or aggregate your orders before you join...
SELECT
Orders.state,
Orders.order_count / StatePopulation.population
FROM
(
SELECT
Orders.state,
COUNT(*) AS order_count
FROM
Orders
GROUP BY
Orders.state
)
Orders
INNER JOIN
StatePopulation
StatePopulation.state = Orders.state
(Please forgive typos and smelling pistakes, I'm doing this on a phone.)
I need to write SQL query like:
Show all countries with more than 1000 users, sorted by user count.
The country with the most users should be at the top.
I have tables:
● Table users (id, email, citizenship_country_id)
● Table countries (id, name, iso)
Users with columns: id, email, citizenship_country_id
Countries with columns: id, name, iso
SELECT countries.name,
Count(users.citiizenship_country_id) AS W1
FROM countries
LEFT JOIN users ON countries.id = users.citizenship_country_id
GROUP BY users.citiizenship_country_id, countries.name
HAVING ((([users].[citiizenship_country_id])>2));
But this does not work - I get an empty result set.
Could you please tell me what I'm doing wrong?
A LEFT JOIN is superfluous for this purpose. To have 1000 users, you need at least one match:
SELECT c.name, Count(*) AS W1
FROM countries c JOIN
users u
ON c.id = u.citizenship_country_id
GROUP BY c.name
HAVING COUNT(*) > 1000;
Notice that table aliases also make the query easier to write and to read.
Group by country name and use HAVING Count(u.citiizenship_country_id)>1000, it filters rows after aggregation:
SELECT c.name,
Count(u.citiizenship_country_id) AS W1
FROM countries c
INNER JOIN users u ON c.id = u.citizenship_country_id
GROUP BY c.name
HAVING Count(u.citiizenship_country_id)>1000
ORDER BY W1 desc --Order top counts first
;
As #GordonLinoff pointed, you can use INNER JOIN instead of LEFT JOIN, because anyway this query does not return counries without users and INNER JOIN performs better because no need to pass not joined records to the aggregation.
In sql server I'm trying to group by each sales people some infos as follow:
I have 2 tables: Positions and Clients
In Positions table, I have the following columns: Client_Id, Balance, Acquisition_Cost and in the Clients table I use the following columns: Client_Id and Sales_person.
I want to group by Sales_person (Clients table) the Client_id, Balance, Acquisition_Cost (Positions table)
I tried this:
SELECT Positions.Client_ID, Positions.Balance, Positions.Acquisition_cost
FROM Positions
INNER JOIN Clients ON Positions.Client_ID = Clients.Client_ID
GROUP BY Sales_person
It gives me "Positions.Client_ID is invalid in the select list because it is not contained in either an aggregate function or the GROUP BY clause".
I precise I'm pretty new on SQL so that does not ring that much a bell to me.
You need to add Positions.Client_ID in the GROUP BY Clause, since the select expects this column if you really do not want Positions.Client_ID in the select, then no need to add in the GROUP BY Clause
For any column in your SELECT that you aren't including in the GROUP BY, you need to use some kind of aggregate function (MAX, SUM, etc.) on the column. So you could write it like this:
SELECT Positions.Client_ID
, Clients.Sales_person
, SUM(Positions.Balance) Balance_Sum,
, SUM(Positions.Acquisition_cost) Acquisition_Cost_Sum
FROM Positions
INNER JOIN Clients ON Positions.Client_ID = Clients.Client_ID
GROUP BY Positions.Client_ID
, Clients.Sales_person
If you only want the totals by Sales_person and not client_ID, you could just remove client_id from the SELECT AND GROUP BY:
SELECT Clients.Sales_person
, SUM(Positions.Balance) Balance_Sum,
, SUM(Positions.Acquisition_cost) Acquisition_Cost_Sum
FROM Positions
INNER JOIN Clients ON Positions.Client_ID = Clients.Client_ID
GROUP BY Clients.Sales_person
SELECT a.Client_ID, b.Sales_person, SUM(a.Balance) as Balance_Sum,SUM(a.Acquisition_cost) as Acquisition_Cost_Sum FROM Positions as a INNER JOIN Clients as b ON a.Client_ID = b.Client_ID GROUP BY a.Client_ID, b.Sales_person
Thanks all of you guys, here's the code that works for me in this case:
SELECT Positions.Client_Id, Clients.Sales_person, SUM(Positions.Balance) as sum_balance_impacted, SUM(Positions.Acquisition_cost) as sum_acquisition_cost
FROM Positions
INNER JOIN Clients ON Positions.Client_Id= Clients.Client_Id
GROUP BY Clients.Sales_person, Positions.Client_Id```
I need to select the sum of transaction columns, count # of transactions, all by distinct customer ids. I have tried a few nested queries as well as something like the following:
select distinct(customer_id), sum(tran_amt), count(tran)
from tran_table
inner join tender_table;
The tender_table has the customer id so I have to join it.
You are using group functions, but you don't have a "group by" clause, I think this will give you the correct answer:
SELECT customer_id, SUM(tran_amt), COUNT(tran)
FROM tran_table INNER JOIN tender_table
GROUP BY CUSTOMER_ID;
I am guessing you want something like this:
select te.customer_id, sum(t.tran_amt), count(*)
from tran_table t inner join
tender_table te
on t.?? = te.??
group by te.customer_id;
You need join conditions and a group by clause. The ?? is for the columns used for the join, which your question does not specify.
Try this :
select tb_tran.customer_id, sum(tran_amt), count(tran)
from tran_table tb_tran inner join tender_table tb_tender
on tb_tran.customer_id = tb_tender.customer_id
group by tb_tran.customer_id;