topic Re: problem with calculating the average using variables defining a range of values in DAX Commands and Tips

problem with calculating the average using variables defining a range of values

Zbig62 — Sun, 21 Apr 2024 11:26:43 GMT

Hello
I am a beginner DAX user and I have a serious problem with calculating the average using variables defining a range of values.
The problem is as follows:

Table: 'public Measurement'

1. average value for the entire range (10 results) as a separate column - works well

AV_N =

CALCULATE (

AVERAGE ( 'public Measurement'[result]);

FILTER (

ALLSELECTED ('public Measurement');

'public Measurement'[characteristicId] = MAX ( 'public Measurement'[characteristicId])

)

The result of the measure is as follows

2. average value for 5 results as a separate column - it seems to work well but the value range is entered "manually"

AV5 =

CALCULATE (

AVERAGEX('public Measurement'; 'public Measurement'[result]);

FILTER (

ALLSELECTED ('public Measurement');

'public Measurement'[characteristicId] = MAX ( 'public Measurement'[characteristicId])

);

'public Measurement'[number] >= 1 ; 'public Measurement'[number] <= 5

)

average value from a range using variables (the measure doesn't work, why?)

Instead of constant values defining the average calculation range (1 and 5 ), I need to use variables.
the variable defining the final range of the value must be calculated, for example:
var stop = count('public measurement'[number]) (instead of the value 5)
How to use variables in such a measure?
Please help

Re: problem with calculating the average using variables defining a range of values

tamerj1 — Sun, 21 Apr 2024 12:19:54 GMT

Hi Zbig62

as part of the debugging process you can create a measure Test = count('public measurement'[number]) and see if returns what you've expected it to return.

Re: problem with calculating the average using variables defining a range of values

Zbig62 — Sun, 21 Apr 2024 12:52:56 GMT

Hello
Thanks for fast reaction.

This is where I have a problem, see the results in the table where the stop_av variable should have the value 10.
As a result, it obtains a bad average value
Please help me, I don't know how to do it

AV =

var stop_av = count('public Measurement'[number])

var start_av = 1

var av =

CALCULATE (

AVERAGEX('public Measurement'; 'public Measurement'[result]);

FILTER (

ALLSELECTED ('public Measurement');

'public Measurement'[characteristicId] = MAX ( 'public Measurement'[characteristicId])

);

'public Measurement'[number] >= start_av ; 'public Measurement'[number] <= stop_av

)

RETURN

Re: problem with calculating the average using variables defining a range of values

Zbig62 — Sun, 21 Apr 2024 16:13:01 GMT

welcome back
Thank you for your suggestions, it allowed me to locate the problem.
Instead of count() I used the countx() function and it works very well
thank you again

var stop_av = CALCULATE(

COUNTX('public Measurement';'public Measurement'[number]);

FILTER(

ALLSELECTED('public Measurement');

'public Measurement'[characteristicId] = MAX('public Measurement'[characteristicId])

)

Re: problem with calculating the average using variables defining a range of values

tamerj1 — Sun, 21 Apr 2024 16:18:19 GMT

Zbig62

Thank you for clarifying your problem one more time. That was clear to me already. My point was that it doesn't seem that you are aware of what var stop_av = count('public Measurement'[number]) returns. You only have an expectation that it should always return 10 while this is not true. In fact this expression will be evaluated within the filter context of evaluation. It n other words it might return different number for each number-characteristicid combination depending how many times this combination is repeated in the Source data table which I personally don't know for lack of information. But I can say it won't be 10 for all combinations unless all combinations have actually 10 rows. If you want to return the maximum [number] for each [caharastaristid] then you can try

var stop_av = CALCULATE ( MAX ( 'public Measurement'[number] ), FILTER ( ALLSELECTED ( 'public Measurement',

'public Measurement'[characteristicId]= MAX ( 'public Measurement'[characteristicId])

)

However, that might not be it as you might face issues with AUTOEXIST. Along with ALLSELECTED, AUTOEXIST might produce problems given that [Result] column is a value column that is part of the filter context.

If you can provide a sample file, it would be much easier to propose a totally different solution perhaps utilizing a NO CALCULATE approach that proofs to he much more reliably in such scenarios.

Re: problem with calculating the average using variables defining a range of values

tamerj1 — Sun, 21 Apr 2024 16:31:58 GMT

Zbig62

There shouldn't be any difference between COUNT and COUNTX. But your new solution made me understand that the number-Characteristicid combination in the source table is unique.

Also COUNT/COUNTX should work properly only if [number] restarts from 1 for each characteristicid.
note: the 'public Measurement'[number] >= start_av a gument is not needed in all cases.