Using ROLLUP in query... Small question - sql

It is a "table_one" in oracle database:
COL1 COL2
-----------------
2 4
2 1
13 14
13 15
I have this query:
SELECT col1, sum(col2) FROM table_one GROUP BY ROLLUP(col1, col2);
After query execute I have:
------------
2 1
2 4
2 5
13 14
13 15
13 29
34
but i need in another way, like this:
------------
2 1
2 4
2 5
13 14
13 15
13 29
without summary of all columns
How I can change my query..... ?

There are two ways to solve this problem. The first is to use grouping sets to define exactly what summary groups you want created.
In this case, you can define to group on (col1) and (col1, col2) usig the following query:
select col1, sum(col2)
from table_one
group by grouping sets ((col1), (col1, col2))
Otherwise, you can group by col1 and rollup on col2 using the following query:
select col1, sum(col2)
from table_one
group by col1, rollup(col2)
Both of these queries should produce the output you require.

Use your current query as a sub query. In the outer query, use a WHERE clause that eliminates the NULL value in Col1
SELECT *
FROM
(
-- Your Rollup Query goes here.
SELECT * FROM MyTable
)
WHERE COL1 IS NOT NULL
A better solution was presented in the comments to eliminate NULL values in the source column(s)
Use the GROUPING function to determine if it is a a sub-total.
http://www.remote-dba.net/pl_sql/t_sql_grouping.htm

Related

Calculating median of 3 columns in a BigQuery table

I am trying to build a query to calculate median of 3 column values. My table looks like below,
Item
Column 1
Column 2
Column 3
A
10
12
4
B
5
14
20
C
15
5
4
I want to be able to output,
Item
Column 1
Column 2
Column 3
Median
A
10
12
4
10
B
5
14
20
14
C
15
5
4
5
I have tried percentile_cont() but that seems to be only for values in a single column. How do i achieve this?
Consider below approach
select *,
( select distinct percentile_disc(col, 0.5) over()
from unnest([Column1, Column2, Column3]) as col
) AS Median
from your_table
if applied to sample data in your question - output is
Have you tried this:
select Col1, Col2, Col3,
PERCENTILE_CONT([Col1, Col2, Col3], 0.5) OVER() AS Median
from tableName

Counting relations between two columns in SQL server

Good morning, I have two columns in sql studio and I need to count the relations between the elements one column and the other one. The problem is it seems only is counting in 'one direction' and I want to know both. Maybe it is easier if I show you in a simple example. I am using SQL server.
This is the original table:
Col1 Col2
3 1
3 2
3 2
4 4
4 5
4 6
3 2
2 3
2 3
And if I do the following (based on count(concat)), it gives the following results.
select Col1, Col2, count(concat(Col1, Col2)) as weight
FROM test1
group by Col1, Col2
Col1 Col2 weight
3 1 1
3 2 3
2 3 2
4 4 1
4 5 1
4 6 1
But what I would like is also counts the relations between the two elements, independently if this is in one column or other. So, the number of relations between 3-2 (or 2-3) would be 5. It is possible to do that?
3 2 5
Any suggestion would be very welcome! Thanks in advance!
Most databases support least() and greatest(). This allows you to do:
select least(col1, col2) as col1, greatest(col1, col2) as col2,
count(*) as cnt
from t
group by least(col1, col2), greatest(col1, col2);
If your database does not support these handy functions, then you can implement similar logic using case expressions.

Finding combinations of specific values

I don't know how to write the query for below.
My table is
col1 col2
5 1
5 5
5 6
5 7
4 5
4 8
4 9
4 3
3 3
3 5
I need to select distinct col1 id where both parameters exists in col2.
eg. if i send 6,7 it should send me 5
This is probably among the fastest solutions:
SELECT col1 -- already DISTINCT?
FROM tbl t1
JOIN tbl t2 USING (col1)
WHERE t1.col2 = 6
AND t2.col2 = 7;
Assuming a PRIMARY KEY or UNIQUE constraint on (col1, col2), like it's typically implemented. Else add DISTINCT.
There are many other ways to implement relational division. Here are some:
How to filter SQL results in a has-many-through relation
Try:
SELECT col1
FROM mytable
WHERE col2 IN (6, 7)
GROUP BY col1
HAVING COUNT(DISTINCT col2) = 2

Return Multiple Rows Based On A Value From Another Column

I have a table that has 3 columns. Col1 is an identity column and sets to autoincrement. Now I want to return multiple rows based on a value in Col3, and the Col2 increments by 1.
For example, if I insert 25 in Col2 and 7 in Col3, I want an output similar to this one:
Col1 Col2 Col3
---- ---- ----
1 25 7
2 26 7
3 27 7
4 28 7
5 29 7
6 30 7
7 31 7
I tried something using partition in my SELECT query, but still I didn't get the desired output. What do you think is the better way to do this?
As far as I understand you're looking for something like this:
SELECT ROW_NUMBER() OVER (ORDER BY col3, col1) AS col2, col3 FROM ...
Provide more details if that doesn't fit.

Multiple row Sums into a Total Column

I have a temp table populated in a sproc that is similar to this:
Company Col1 Col2 Col3 Total
Company1 4 3 2 9
Company2 1 0 3 4
Total ? ? ? ?
Is there a slick way to get the bottom total column populated with SUM of each row in one shot without having to do each column individually using sub-queries?
select sum(col1), sum(col2), sum(col3), sum(col1+col2+col3)
FROM CompanyTable
If the DBMS you use is MS SQL than be aware of how to reference the temporary table:
SELECT SUM(Col1) as TotalOfCol1,...
FROM #Company
If you are using MySQL, you can get the whole table in one query using the WITH ROLLUP feature:
SELECT Company, SUM(Col1), SUM(Col2), SUM(Col3), SUM(Col1 + Col2 + Col3)
FROM Table1
GROUP BY Company
WITH ROLLUP
Results:
'Company1' 4 3 2 9
'Company2' 1 0 3 4
'' 5 3 5 13 <--- This extra row is the column totals