MySQL extract average data from multiple group criteria - mysql

Question

Apologies for the wall of text, the example in the end explains my question. Any help is appreciated, thank you!
I have a table which contains several columns of data from among other values voltages and currents.
These instances are logged every second when there is current flowing. I want to calculate an approximated kJ and kW from these values.
Basically I have one table, instances, that contains:
instanceID,
location,
current,
voltage,
time.
And another one, sets, that contains:
instanceID,
setID.
The instanceID is the same, the instanceID in instances is a FK pointing to sets. For every location in instances there are approximately 23 rows (varies). There are 30 locations. So I have 23 rows where the instance have location 1, another 23 for the same instance for location 2 and so on. Time is the logged time for when the measured data was taken (so if the difference is one second between all the 23 instances the difference between the first and the last time is 23 seconds).
I need to calculate the average kW and the total kJ (approximated).
What I've done is the following:
SELECT instances.instanceID, location, current,
voltage, current * voltage AS kW,
COUNT(IF(current > 0 AND voltage > 0,
instances.instanceID,
0)) AS InstancedTime
FROM instances
INNER JOIN sets ON instances.instanceID = sets.instanceID
WHERE sets.setID = arbitrary_number;
The problem arises that I get the following table:
instanceID, location, current, voltage, kW, InstancedTime
The kW is a random number from one of the 23 sets, which is fine since it's an approximation anyway, but the COUNT(IF()) is counting ALL the instances in the instances table, when I only want the query to count the instances for every location.
I tried the MAX(CAST(time AS SIGNED)) - MIN(CAST(time AS SIGNED)), but that takes the max time from the last location minus the min time of the first location, I want to isolate it to one location at a time.
What I want to do is get the total amount of kJ which would be the time it had power multiplied by the kW of that time. Since I know the time is always 1 second between the instances it should be enough to count the number of instances for individual locations and multiply that by the kW, however I want to do that for all the instances within one set. It is possible to replace the set by using a single query for all the individual instances but that would take eons.
I'm trying to take a table that looks like
instanceID, location, current, voltage, kW, InstancedTime
1, 1, 500V, 2A, 1kW, 1s
1, 1, 500V, 2A, 1kw, 1s
1, 2, 400V, 3A, 1.2kW, 1s
1, 2, 400V, 3A, 1.2kW, 1s
2, 1, 700V, 2A, 1.4kW, 1s
2, 1, 700V, 2A, 1.4kw, 1s
2, 2, 300V, 3A, 0.9kW, 1s
2, 2, 300V, 3A, 0.9kW, 1s
And add the kJ which would be summarising the number of instances that ID 1 has been in location 1 and location 2, doing the same for ID 2 and presenting this all in one table that would look like:
instanceID, location, current, voltage, kW, SumInstancedTime, kJ
1, 1, 500V, 2A, 1kW, 2s, 2kJ
1, 2, 400V, 3A, 1.2kW, 2s, 2.4kJ
2, 1, 700V, 2A, 1.4kW, 2s, 2.8kJ
2, 2, 300V, 3A, 0.9kW, 2s, 1.8kJ
Thank you for your time, any provided help is appreciated!

Answer 1

I cannot test my answer right now, but it sounds that what you need is a GROUP BY:
the following query is an example that averages your current and voltage for every set of instance/location and then calculates the value
SELECT instances.instanceID, location, AVG(current) as avg_current,
AVG(voltage) as avg_voltage, AVG(current) * AVG(voltage) AS kW,
COUNT(IF(current > 0 AND voltage > 0,
instances.instanceID,
0)) AS InstancedTime
FROM instances
INNER JOIN sets ON instances.instanceID = sets.instanceID
WHERE sets.setID = arbitrary_number;
GROUP BY instances.instanceID, location

Answer 2

This is an example where you are trying to group consecutive rows in a table. In your example, they are not interleaved, but I'm assuming they could be. You need to assign everything with the same instanceID and location to the same group.
My approach is to find the next higher instance/location pair, and to assign that as a group identifier. I do this using a subquery. Once I have this identifier, I just summarize each group:
select i.instanceId, i.location, i.current, i.volate, i.kw, COUNT(*) as SumTime,
SUM(kw) as KJ
from (select i.*,
(select concat(i2.instanceId, ',', i2.location)
from instances i2
where i2.instanceId > i.instanceId or (i2.instanceId = i.instanceId and i2.location > i.location)
order by i2.instanceId desc, i2.locationId desc
limit 1
) as grouping
from instances i
) i
group by grouping