I have a large data set with about 100 million rows that I want to 'compress' the data set and get a 1% sample of the entire dataset while ensuring relativity.
How can such query be implemented?
Step 1: create the helper table
You can use aggregation to group records by visit_id, and CROSS JOIN with a query that computes the total number of records in the table to compute the distribution percent:
CREATE TABLE my_helper AS
SELECT
t.visit_number,
COUNT(*) visit_count,
SUM(t.purchase_id) sum_purchase,
COUNT(*)/total.cnt distribution
FROM
mytable t
CROSS JOIN (SELECT COUNT(*) cnt FROM mytable) total
GROUP BY t.visit_number
Step 2: sample the main table using the helper table
Within a subquery, you can use ROW_NUMBER() OVER(PARTITION BY visit_number ORDER BY RANDOM()) to assign a random rank to each record within groups of records sharing the same visit_id. Then, in the outer query, you can join on the helper table to select the corect amount of records for each visit_id:
SELECT x.*
FROM (
SELECT
t.*,
ROW_NUMBER() OVER(PARTITION BY visit_number ORDER BY RANDOM()) rn
FROM mytable t
) x
INNER JOIN my_helper h ON h.visit_number = x.visit_number
WHERE x.rn <= 1000000 * h.distribution
Side notes:
this only works if there are indeed more than 1 million record in the source table
the exact number of records in the output might be slightly below or above 1 million (depending on the distribution in the original table)
it should be possible to combine both queries into a single one, which would avoid the need to use a helper table
This is doable. A quick way is to take every nth record only.
1) order by a random column (probably ID)
2) apply a nownum() attribute
3) apply a mod(rownum) = 0 on whatever percent makes sense (e.g. 1% would be rownum mod 100)
You may need steps 1/2 in a sub query and step 3 on the outside.
Enjoy and good luck!
I want to write a SQL query to generate ranks of groups and subgroups based on third column (Price in this case). While i know we can use dense_rank() to generate ranks based on one column. I have no idea how to generate the two columns of ranks as shown below in a single query.
Both the rankings are based on price. So J3 comes first because J3 sum(price) is 1600. J1 comes second because J1 sum(price) is 1500 and so on.
Any inputs are appreciated.
I have provided the sample input and output. The name of the input table is "RENTAL"
First roll up jet_type prices to the jet_type level, then create a ranking of all jet_types ordered by rolled up price, and finally use your window function in the outer query partitioned by jet_price and ordered by highest rolled up price to create rank_service_wthin_jet:
select a.jet_type, b.rownum rank_jet, a.service_type, a.price,
row_number() over(partition by a.jet_type order by a.price desc) rank_service_wthin_jet
from yourtable a join (
select jet_type, row_number() over(order by price desc) rownum from (
select jet_type, sum(price) price from yourtable
group by jet_type)a)b on a.jet_type=b.jet_type
You can generate two columns as:
select t.*,
dense_rank() over (order by jet_type) as rank_jet,
row_number() over (partition by jet_type order by price desc) as rank_service_within_jet
. . .
This does not exactly return what is in your table. But the results are quite similar and -- even more important -- make sense.
i have a tbl call it tbl_test in which continously data are inserting and it has approx 10^6 records at a time it has colums
Acquire_Id (Value between 1 to 20 ),
Status_Msg(value between 'A' to 'Z'),
Status_Code(value between 1 to 26)
There is one to one mapping b/w Status_Msg and Status_Code
Now i want to get two most frequent status_msg and Staus_Code Count for each acquirer if they are present in table
Query should be Cost Saving
Most databases support the ANSI standard window functions. You can get what you want using row_number() (or rank() or dense_rank(), depending on how ties are returned) after aggregating the values.
The following returns exactly two rows for each acquirer (even if there are ties).
select t.*
from (select t.acquire_id, t.status_msg, t.status_code, count(*) as cnt,
row_number() over (partition by t.acquire_id order by count(*) desc) as seqnum
from tbl_test t
group by t.acquire_id, t.status_msg, t.status_code
) t
where seqnum <= 2;
I'm looking to pull the top 10% of a summed value on a Postgres sever.
So i'm summing a value with sum(transaction.value) and i'd like the top 10% of the value
From what I gather in your comments, I assume you want to:
Sum transactions per customer to get a total per customer.
List the top 10 % of customers who actually have transactions and spent the most.
WITH cte AS (
SELECT t.customer_id, sum(t.value) AS sum_value
FROM transaction t
GROUP BY 1
)
SELECT *, rank() OVER (ORDER BY sum_value DESC) AS sails_rank
FROM cte
ORDER BY sum_value DESC
LIMIT (SELECT count(*)/10 FROM cte)
Major points
Best to use a CTE here, makes the count cheaper.
The JOIN between customer and transaction automatically excludes customers without transaction. I am assuming relational integrity here (fk constraint on customer_id).
Dividing bigint / int effectively truncates the result (round down to the nearest integer). You may be interested in this related question:
PostgreSQL equivalent for TOP n WITH TIES: LIMIT "with ties"?
I added a sails_rank column which you didn't ask for, but seems to fit your requirement.
As you can see, I didn't even include the table customer in the query. Assuming you have a foreign key constraint on customer_id, that would be redundant (and slower). If you want additional columns from customer in the result, join customer to the result of above query:
WITH cte AS (
SELECT t.customer_id, sum(t.value) AS sum_value
FROM transaction t
GROUP BY 1
)
SELECT c.customer_id, c.name, sub.sum_value, sub.sails_rank
FROM (
SELECT *, rank() OVER (ORDER BY sum_value DESC) AS sails_rank
FROM cte
ORDER BY sum_value DESC
LIMIT (SELECT count(*)/10 FROM cte)
) sub
JOIN customer c USING (customer_id);
I want to use the ROW_NUMBER() to get...
To get the max(ROW_NUMBER()) --> Or i guess this would also be the count of all rows
I tried doing:
SELECT max(ROW_NUMBER() OVER(ORDER BY UserId)) FROM Users
but it didn't seem to work...
To get ROW_NUMBER() using a given piece of information, ie. if I have a name and I want to know what row the name came from.
I assume it would be something similar to what I tried for #1
SELECT ROW_NUMBER() OVER(ORDER BY UserId) From Users WHERE UserName='Joe'
but this didn't work either...
Any Ideas?
For the first question, why not just use?
SELECT COUNT(*) FROM myTable
to get the count.
And for the second question, the primary key of the row is what should be used to identify a particular row. Don't try and use the row number for that.
If you returned Row_Number() in your main query,
SELECT ROW_NUMBER() OVER (Order by Id) AS RowNumber, Field1, Field2, Field3
FROM User
Then when you want to go 5 rows back then you can take the current row number and use the following query to determine the row with currentrow -5
SELECT us.Id
FROM (SELECT ROW_NUMBER() OVER (ORDER BY id) AS Row, Id
FROM User ) us
WHERE Row = CurrentRow - 5
Though I agree with others that you could use count() to get the total number of rows, here is how you can use the row_count():
To get the total no of rows:
with temp as (
select row_number() over (order by id) as rownum
from table_name
)
select max(rownum) from temp
To get the row numbers where name is Matt:
with temp as (
select name, row_number() over (order by id) as rownum
from table_name
)
select rownum from temp where name like 'Matt'
You can further use min(rownum) or max(rownum) to get the first or last row for Matt respectively.
These were very simple implementations of row_number(). You can use it for more complex grouping. Check out my response on Advanced grouping without using a sub query
If you need to return the table's total row count, you can use an alternative way to the SELECT COUNT(*) statement.
Because SELECT COUNT(*) makes a full table scan to return the row count, it can take very long time for a large table. You can use the sysindexes system table instead in this case. There is a ROWS column that contains the total row count for each table in your database. You can use the following select statement:
SELECT rows FROM sysindexes WHERE id = OBJECT_ID('table_name') AND indid < 2
This will drastically reduce the time your query takes.
You can use this for get first record where has clause
SELECT TOP(1) * , ROW_NUMBER() OVER(ORDER BY UserId) AS rownum
FROM Users
WHERE UserName = 'Joe'
ORDER BY rownum ASC
ROW_NUMBER() returns a unique number for each row starting with 1. You can easily use this by simply writing:
ROW_NUMBER() OVER (ORDER BY 'Column_Name' DESC) as ROW_NUMBER
May not be related to the question here. But I found it could be useful when using ROW_NUMBER -
SELECT *,
ROW_NUMBER() OVER (ORDER BY (SELECT 100)) AS Any_ID
FROM #Any_Table
select
Ml.Hid,
ml.blockid,
row_number() over (partition by ml.blockid order by Ml.Hid desc) as rownumber,
H.HNAME
from MIT_LeadBechmarkHamletwise ML
join [MT.HAMLE] h on ML.Hid=h.HID
SELECT num, UserName FROM
(SELECT UserName, ROW_NUMBER() OVER(ORDER BY UserId) AS num
From Users) AS numbered
WHERE UserName='Joe'
You can use Row_Number for limit query result.
Example:
SELECT * FROM (
select row_number() OVER (order by createtime desc) AS ROWINDEX,*
from TABLENAME ) TB
WHERE TB.ROWINDEX between 0 and 10
--
With above query, I will get PAGE 1 of results from TABLENAME.
If you absolutely want to use ROW_NUMBER for this (instead of count(*)) you can always use:
SELECT TOP 1 ROW_NUMBER() OVER (ORDER BY Id)
FROM USERS
ORDER BY ROW_NUMBER() OVER (ORDER BY Id) DESC
Need to create virtual table by using WITH table AS, which is mention in given Query.
By using this virtual table, you can perform CRUD operation w.r.t row_number.
QUERY:
WITH table AS
-
(SELECT row_number() OVER(ORDER BY UserId) rn, * FROM Users)
-
SELECT * FROM table WHERE UserName='Joe'
-
You can use INSERT, UPDATE or DELETE in last sentence by in spite of SELECT.
SQL Row_Number() function is to sort and assign an order number to data rows in related record set. So it is used to number rows, for example to identify the top 10 rows which have the highest order amount or identify the order of each customer which is the highest amount, etc.
If you want to sort the dataset and number each row by seperating them into categories we use Row_Number() with Partition By clause. For example, sorting orders of each customer within itself where the dataset contains all orders, etc.
SELECT
SalesOrderNumber,
CustomerId,
SubTotal,
ROW_NUMBER() OVER (PARTITION BY CustomerId ORDER BY SubTotal DESC) rn
FROM Sales.SalesOrderHeader
But as I understand you want to calculate the number of rows of grouped by a column. To visualize the requirement, if you want to see the count of all orders of the related customer as a seperate column besides order info, you can use COUNT() aggregation function with Partition By clause
For example,
SELECT
SalesOrderNumber,
CustomerId,
COUNT(*) OVER (PARTITION BY CustomerId) CustomerOrderCount
FROM Sales.SalesOrderHeader
This query:
SELECT ROW_NUMBER() OVER(ORDER BY UserId) From Users WHERE UserName='Joe'
will return all rows where the UserName is 'Joe' UNLESS you have no UserName='Joe'
They will be listed in order of UserID and the row_number field will start with 1 and increment however many rows contain UserName='Joe'
If it does not work for you then your WHERE command has an issue OR there is no UserID in the table. Check spelling for both fields UserID and UserName.