My data set is a list of house sales. I am trying to update the address column so that any NULL values get replaced with an actual address. I am getting these replacement addresses from other inputs within the data. There are other sales of the same house so they share a "parcelid". I am doing an UPDATE with COALESCE and an INNER JOIN to complete the task. I am doing so with this code.
UPDATE house
SET property_address =
COALESCE(house1.property_address,house2.property_address)
FROM house AS house1
JOIN house AS house2
ON house1.parcelid = house2.parcelid
AND house1.uniqueid != house2.uniqueid
WHERE house1.property_address IS NULL
When I look at my 'house' table though, every single entry in the 'property_address' line has been updated to the same address. When I did some digging that address is the first address that occurs when I look at the join with this code.
SELECT
house1.parcelid,
house2.parcelid,
house1.property_address,
house2.property_address,
COALESCE(house1.property_address,house2.property_address)
FROM house AS house1
JOIN house AS house2
ON house1.parcelid = house2.parcelid
AND house1.uniqueid != house2.uniqueid
WHERE house1.property_address IS NULL
How to I get it to one ignore any addresses that are not null and two update to the appropriate address and not just keep copying the same address. Thanks for the help. (This is my first time asking a question here so any pointers on how to properly format would be appreciated)
Related
I am looking for a query that separate the data with the condition WHERE in the same output but in separates columns.
Example: I have the table Product_2:
I have two separates queries (to separate the products by Produt_Tag):
SELECT
Product_Mark AS "PIT-10_Product_Mark",
Product_Model AS "PIT-10_Product_Model"
FROM Product_2
WHERE Product_Tag = 'PIT-10';
SELECT
Product_Mark AS "PIT-11_Product_Mark",
Product_Model AS "PIT-11_Product_Model"
FROM Product_2
WHERE Product_Tag = 'PIT-11';
And I get this output:
But I need the output to be like this:
Can someone tell me how I need to modify my query to have the four columns in the same table/ output?
Thank you
I forgot to tell that in the data I Have the “Porduct_Mark” that only appears one time. (in reality the data in “Product_Mark” is the name of the place where the instrument is located and one place can have one or two instruments “Product_Model”. At the end I’m looking for the result show in the image here below. I tried to use LEFT JOIN but that don’t work.
here is the new table "Product_2"
Result that I'm looking for:
Luis Ardila
I am assuming Product_PK is the primary key for the table and the repeated value 1002 shown in the question is a mistake. Considering this assumption, you can get the result set using self join as below.
SELECT pa.Product_Mark AS "PIT-10_Product_Mark", pa.Product_Model AS "PIT-10_Product_Model",
pb.Product_Mark AS "PIT-11_Product_Mark", pb.Product_Model AS "PIT-11_Product_Model"
FROM Product_2 pa
INNER JOIN Product_2 pb
ON pa.Product_Mark = pb.Product_Mark
WHERE pa.product_pk != pb.product_pk
and pa.Product_Tag = 'PIT-10'
and pb.Product_Tag = 'PIT-11';
verified same in https://dbfiddle.uk/NiOO8zc1
I have 2 tables, 1 containing the main body of information, the second contains information on country naming convensions. in the information table, countries are identified by Name, I would like to update this string to contain an ISO alpha 3 value which is contained in the naming convention table. e.g turning "United Kingdom" -> "GBR"
I have wrote the following query to make the update, but it effects 0 rows
UPDATE
`db.catagory.test_votes_ds`
SET
`db.catagory.test_votes_ds`.country = `db.catagory.ISO-Alpha`.Alpha_3_code
FROM
`db.catagory.ISO-Alpha`
WHERE
`LOWER(db.catagory.ISO-Alpha`.Country) = LOWER(`db.catagory.test_votes_ds`.country)
I've done an inner join outside of the update between the 2 to make sure that the values are compatable and it returns the correct value, any ideas as to why it isn't updating?
The join used to validate the result is listed below, along with the result:
SELECT
`db.catagory.test_votes_ds`.country, `db.catagory.ISO-Alpha`.Alpha_3_code
from
`db.catagory.test_votes_ds`
inner join
`db.catagory.ISO-Alpha`
on
LOWER(`db.catagory.test_votes_ds`.country) = LOWER(`db.catagory.ISO-Alpha`.Country)
1,Ireland,IRL
2,Australia,AUS
3,United States,USA
4,United Kingdom,GBR
This is not exactly an answer. But your test may not be sufficient. You need to check where the values do not match. So, to return those:
select tv.*
from `db.catagory.test_votes_ds` tv left join
`db.catagory.ISO-Alpha` a
on LOWER(tv.country) = LOWER(a.Country)
where a.Country IS NULL;
I suspect that you will find countries that do not match. So when you run the update, the matches are getting changed the first time. Then the non-matches are never changed.
i want to group this query by project only because there are two records of same project but i only want one.
But when i add group by clause it asks me to add other columns as well by which grouping does not work.
*
DECLARE #Year varchar(75) = '2018'
DECLARE #von DateTime = '1.09.2018'
DECLARE #bis DateTime = '30.09.2018'
select new_projekt ,new_geschftsartname, new_mitarbeitername, new_stundensatz
from Filterednew_projektkondition ps
left join Filterednew_fakturierungsplan fp on ps.new_projekt = fp.new_hauptprojekt1
where ps.statecodename = 'Aktiv'
and fp.new_startdatum >= #von +'00:00:00'
and fp.new_enddatum <= #bis +'23:59:59'
--and new_projekt= Filterednew_projekt.new_
--group by new_projekt
*
look at the column new_projekt . row 2 and 3 has same project, but i want it to appear only once. Due to different other columns this is not possible.
if its of interested , there is another coluim projectcondition id which is unique for both
You can't ask a database to decide arbitrarily for you, which records should be thrown away when doing a group. You have to be precise and specific
Example, here is some data about a person:
Name, AddressZipCode
John Doe, 90210
John Doe, 12345
SELECT name, addresszipcode FROM person INNER JOIN address on address.personid = person.id
There are two addresses stored for this one guy, the person data is repeated in the output!
"I don't want that. I only want to see one line for this guy, together with his address"
Which address?
That's what you have to tell the database
"Well, obviously his current address"
And how do you denote that an address is current?
"It's the one with the null enddate"
SELECT name, addresszipcode FROM person INNER JOIN address on address.personid = person.id WHERE address.enddate = null
If you still get two addresses out, there are two address records that are null - you have data that is in violation of your business data modelling principles ("a person's address history shall have at most one adddress that is current, denoted by a null end date") - fix the data
"Why can't i just group by name?"
You can, but if you do, you still have to tell the database how to accumulate the non-name data that it shows you. You want an address data out of it, it has 2 it wants to show you, you have to tell it which to discard. You could do this:
SELECT name, MAX(addresszipcode) FROM person INNER JOIN address on address.personid = person.id GROUP BY name
"But I don't want the max zipcode? That doesn't make sense"
OK, use the MIN, the SUM, the AVG, anything that makes sense. If none of these make sense, then use something that does, like the address line that has the highest end date, or the lowest end date that is a future end date. If you only want one address on show you must decide how to boil that data down to just one record - you have to write the rule for the database to follow and no question about it you have to create a rule so make it a rule that describes what you actually want
Ok, so you created a rule - you want only the rows with the minimum new_stundenstatz
DECLARE #Year varchar(75) = '2018'
DECLARE #von DateTime = '1.09.2018'
DECLARE #bis DateTime = '30.09.2018'
select new_projekt ,new_geschftsartname, new_mitarbeitername, new_stundensatz
from
(SELECT *, ROW_NUMBER() OVER(PARTITON BY new_projekt ORDER BY new_stundensatz) rown FROM Filterednew_projektkondition) ps
left join
Filterednew_fakturierungsplan fp on ps.new_projekt = fp.new_hauptprojekt1
where ps.statecodename = 'Aktiv'
and fp.new_startdatum >= #von +'00:00:00'
and fp.new_enddatum <= #bis +'23:59:59'
and ps.rown = 1
Here I've used an analytic operation to number the rows in your PS table. They're numbered in order of ascending new_stundensatz, starting with 1. The numbering restarts when the new_projekt changes, so each new_projekt will have a number 1 row.. and then we make that a condition of the where
(Helpful side note for applying this technique in future.. Ff it were the FP table we were adding a row number to, we would need to put AND fp.rown= 1 in the ON clause, not the WHERE clause, because putting it in the where would make the LEFT join behave like an INNER, hiding rows that don't have any FP matching record)
So Here is the problem I have a requirement where I need a customer type to equal two different things.
To Cover the requirement I don't need the customer type to equal Client, or Non client but equal Client, and Non_Client. Each Customer_No can have multiple Customer Types
Here is an example of what I have worked on so far. If you know a better way of optimizing this as well as solving the problem please let me know.
The out put should look like this
CustomerID CustomerType CustomerType
--------------------------------------
2345 Client NonClient
Select TB1.Customer_ID, IB1.Customer_Type, AS Non_client IB1.Customer_Type AS Client
From Client TB1, Client_ReF XB1, Client_Instr IB1, Client_XREC FB1
Where XB1.Client_NO = TB1.Client_NO
AND FB1.Client_ACCT = TB1.ACCT
AND XB1.Client_Instruct_NO = IB1.Client_Instruct_NO
AND FB1.Customer_ID= TB1. Client_NO
AND IB1.Client = 'Client'
AND IB1.Non_Client = 'NonClient'
I have omitted a few other filters that I felt were unnecessary. This also may not make sense, but I tried to change up the names of stuff as to keep myself in compliance.
First a small syntactic error:
You mustn't have a comma before the "AS Non_client "
Then what you are trying to do is make 1 value equal 2 different things for the same column which can never be true:
IB1.Customer_Type for 1 record can never be equal to "Client" and "NonClient" simultaneously.
The key here is that 1 customer can have multiple records and the records can differ in the customer_type. So to use that we need to join those records together which is easy since they share a Customer_ID:
Select TB1.Customer_ID,
IB1.Customer_Type AS Client,
IB2.Customer_Type AS Non_client
From Client TB1,
Client_ReF XB1,
Client_Instr IB1,
Client_Instr IB2,
Client_XREC FB1
Where XB1.Client_NO = TB1.Client_NO
AND FB1.Client_ACCT = TB1.ACCT
AND XB1.Client_Instruct_NO = IB1.Client_Instruct_NO
AND FB1.Customer_ID= TB1.Client_NO
AND IB1.Client = 'Client'
AND XB1.Client_Instruct_NO = IB2.Client_Instruct_NO
AND IB2.Non_Client = 'NonClient';
The above may not actually work due to me not fully understanding your data and structures but should put you on the right path. Particularly around the join of IB2 with XB1, you might have to join IB2 with all the same tables as IB1.
A better way than that however, and i'll leave you to research it, is using the EXISTS statement. The difference is that the above will join all records for the same customer together whereas EXISTS will just be satisfied if there's at least 1 instance of a "NonClient" record.
I have a many to many relationship between people and some electronic codes. The table with the codes has the code itself, and a text description of the code. A typical result set from a query might be (there are many codes that contain "broken" so I feel like it's better to search the text description rather than add a bunch of ORs for every code.)
id# text of code
1234 broken laptop
1234 broken mouse
Currently the best way for me to get a result set like this is to use the LIKE%broken% filter. Without changing the text description, is there any way I can return only one instance of a code with broken? So in the example above the query would only return 1234 and broken mouse OR broken laptop. In this scenario it doesn't matter which is returned, all I'm looking for is the presence of "broken" in any of the text descriptions of that person's codes.
My solution at the moment is to create a view that would return
`id# text of code
1234 broken laptop
1234 broken mouse`
and using SELECT DISTINCT ID# while querying the view to get only one instance of each.
EDIT ACTUALLY QUERY
SELECT tblVisits.kha_id, tblICD.descrip, min(tblICD.Descrip) as expr1
FROM tblVisits inner join
icd_jxn on tblVisits.kha_id = icd_jxn.kha)id inner join tblICD.icd_fk=tblICD.ICD_ID
group by tblVisits.kha_id, tblicd.descrip
having (tblICD.descrip like n'%broken%')
You could use the below query to SELECT the MIN code. This will ensure only text per id.
SELECT t.id, MIN(t.textofcode) as textofcode
FROM table t
WHERE t.textofcode LIKE '%broken%'
GROUP BY t.id
Updated Actual Query:
SELECT tblVisits.kha_id,
MIN(tblICD.Descrip)
FROM tblVisits
INNER JOIN icd_jxn ON tblVisits.kha_id = icd_jxn.kha)id
INNER JOIN tblicd ON icd_jxn.icd_fk = tbl.icd_id
WHERE tblICD.descrip like n'%broken%'
GROUP BY tblVisits.kha_id