Separating/Sorting single column values into several columns using case function - mysql

Question

I have two tables that I want to join and split with a case function depending on the values in one of the columns. (I know, sounds weird so let me explain)
It's a process where I run separate batches. Every batch has several samples that are measured in instances of voltage readings in several locations. My two tables looks like this:
Sample Readings
id id
BatchesID SampleID
... voltage
... location
When a batch is run, it takes one sample at a time and for every location (25 locations) it takes about 20 readings of the voltage before moving on to the next one.
I want to look at one batch at a time, and for every Sample.id, I want to gather the AVG(voltage) for all the locations. My table for Readings turns out like:
SampleID location voltage
1 1 5.23
1 1 4.53
... ... ...
1 25 7.89
2 1 4.96
2 1 5.04
... ... ...
2 25 6.09
...
But I want it to look like:
SampleID avg_v_for_1 avg_v_for_2 ... avg_v_for_25
1 4.73 5.24 ... 6.35
2 3.87 4.76 ... 9.32
... ... ... ... ...
200 6.73 3.87 ... 8.23
Basically, what I want to do is for every separate sample, I want to take the average voltage for all the measurements in every location and put in on a single row. What my current syntax looks like is this:
SELECT Readings.SampleID, Sample.BatchesID
(case when location = '1' then AVG(voltage) else 0 end) avg_v_for_1,
(case when location = '2' then AVG(voltage) else 0 end) avg_v_for_2,
...
(case when location = '25' then AVG(voltage) else 0 end) avg_v_for_25
FROM DB.Readings
INNER JOIN Sample
ON Readings.SampleID = Sample.id
WHERE Sample.BatchesID = 'specific_batch_id'
GROUP BY Readings.location, Sample.id;
The problem is that this generates the following table:
SampleID avg_v_for_1 avg_v_for_2 ... avg_v_for_25
1 4.73 0 ... 0
1 0 4.76 ... 0
1 0 0 ... 6.73
2 3.87 0 ... 0
2 0 4.83 ... 0
...
How can I get MySQL to gather ALL the average values for EVERY location on a SINGLE row? I have tried removing the group by location and only group by sampleID but then I only get the values for the first location and everything else becomes 0.
Any help is appreciated, thank you!

Answer 1

I add another answer with explanation how the the query with AVG(case ..when ... then..end) works, and why the version with case ... when ... then AVG(..) end doesn't give expected results.
The first remark: the ANSI SQL standard for group by queries is the following:
SELECT column1, column2, ... column_n, aggregate_function (expression)
FROM tables
WHERE predicates
GROUP BY column1, column2, ... column_n;
where aggregated_function can be a function such a: SUM, MAX, MIN, COUNT, AVG
There are several rules (restrictions) for the GROUP BY CLASUE, see this link for details: http://etutorials.org/SQL/Mastering+Oracle+SQL/Chapter+4.+Group+Operations/4.2+The+GROUP+BY+Clause/
one of them says that:
GROUP BY clause must include all nonaggregate expressions
It means, that all columns in SELECT clause must be listed in the GROUP BY clause,
for example this query:
SELECT col1, col2, AVG( expression )
FROM table
GROUP BY col2
is wrong, because col1 is not listed in the GROUP BY clause, and this query won't work on all databases (Oracle, Postgresql, MS-SQL etc.) - except MySql (why - I'll tell about it later).
The expression within the aggregated function can refer to all columns of the table, regardless of the column is listed in the GROUP BY clause or not.
Because of the above the query:
SELECT Readings.SampleID,
(case when location = '1' then AVG(voltage) else 0 end) avg_v_for_1
....
GROUP BY sampleId
simply won't work on all databases that are compliant with ANSI SQL, this query will give a syntax error because location is out of AVG function, but is not listed in the GROUP BY clause.
The question - why this query works on MySql ?
Because MySql implemented it's own extension to the GROUP BY query, see this link --> http://dev.mysql.com/doc/refman/5.6/en/group-by-extensions.html
In MySql the select list can refer to nonaggregated columns not listed in the GROUP BY clause. Becaue of this extension our query is syntactically correct and runs on MySql, but gives unexpected (unwanted) results, since an order of expression's evaluation is different:
1. it first runs an aggregated (group by) query and evaluates AVG( price ),
2. then evaluates CASE WHEN ... THEN, but for resultset returned by the aggregated query from point 1
The query with the clause AVG( case when ... then ):
1. first calucates the expression CASE-WHEN-THEN for all table rows
2. then runs an aggregated query for resultset returned by #1 and calculates the AVG.

Answer 2

Try:
SELECT Readings.SampleID, Sample.BatchesID
AVG(case when location = '1' then voltage else null end) avg_v_for_1,
AVG(case when location = '2' then voltage else null end) avg_v_for_2,
...
AVG(case when location = '25' then voltage else null end) avg_v_for_25
FROM DB.Readings
........
GROUP BY sample_id
--- EDIT --> use ifnull function to change nulls into 0
SELECT Readings.SampleID, Sample.BatchesID
ifnull( AVG(case when location = '1' then voltage else null end), 0 ) avg_v_for_1,
ifnull( AVG(case when location = '2' then voltage else null end), 0 ) avg_v_for_2,
...
ifnull( AVG(case when location = '25' then voltage else null end), 0 ) avg_v_for_25
FROM DB.Readings
........
GROUP BY sample_id