Alternative for AverageX

[Anonymous]

Hi, I have following working dax code. Since the data volume is high, it runs slow and gives memory error when filtered to lowest level. Is there a better alternative to using @AverageX?

The code is:

Average Balance (Period) = AVERAGEX( filter ( ALLSELECTED('Date'[Date])|and('Date'[Date]>=min('Date'[Date])|'Date'[Date] <= max ('Date'[Date])))| [Daily Average Balance] )

where,

Daily Average Balance = if (ISBLANK([(Cumulative) Opening Balance])| [(Cumulative) Closing Balance]| ([(Cumulative) Closing Balance]+[(Cumulative) Opening Balance])/2 )

Since [Daily Average Balance], [(Cumulative) Closing Balance] & [(Cumulative) Opening Balance] all are measures, I am unable to use Average function.

Thanks in advance!

Shailee.

[sturlaws]

Hi @Anonymous 

could you try to rewrite the code to this:

Average Balance (Period) =
AVERAGEX (
    values(account number),
    [Daily Average Balance]
)

The idea is that the daily average balance has to be computed for each account number, and after that compute the average balance for all accounts. There is no need to include any time handling in the expression, it will be evaluated in the context from the report. E.g. if the report has a month slicer and April is selected, then the measure will calculate the average balance of April 

Cheers,
Sturla

[Anonymous]

@sturlaws 

Though your suggestion improves the performance, but it does not yield the desired results. It does not consider the dates when there was no transaction for a particular Accound Code. Also I have multiple hierarchy levels for Account code, and report can be drawn at any level.

[sturlaws]

hm, 

not sure if this will yield any better performance:

Average Balance (Period) =
VAR _minDate =
    MIN ( 'Date'[Date] )
VAR _maxDate =
    MAX ( 'Date'[Date] )
RETURN
    AVERAGEX (
        FILTER ( ALL ( 'Date' ), 'Date'[Date] >= _minDate && 'Date'[Date] <= _maxDate ),
        [Daily Average Balance]
    )

If you provide some sample data it will be easier to help you

[Anonymous]

@sturlaws 

There was not a major performance improvement with your last suggestion.

Following is relevant sample dataset-

This is the Accounting table on which Average Balance (rolling) needs to be calculated on [Daily Average Balance], which is a measure based on Home_Currency_Amount, Debit_Credit, and related Date and GL Master tables. This measure works perfectly fine separately.

source_system_code	entity_code	effective_date	posted_date	SubLob_Code	GL_code	home_currency_amount	debit_credit
ORA01	181	04/01/2020	04/01/2020	183	727104	213	DR
ORA01	101	04/01/2020	04/01/2020	G41	733102	1269.82	DR
ORA01	101	04/01/2020	04/01/2020	G41	733102	1006.5	CR
ORA01	181	04/01/2020	04/01/2020	E92	734113	303	CR
ORA01	181	05/01/2020	05/01/2020	183	727102	66	DR

The Accounting table is linked to a Date Table through Effective Date and following GL Master table (linked through SourceSystemCode + GL Code):

Source_System_Code	GL_CODE_DESCRIPTION	GL Code Level 1	GL Code Level 2	GL Level 3	ACCOUNT_TYPE
IMB01	KRISHNA PRATAP SINGH	KRISHNA PRATAP SINGH	Trade payables to non-related parties	Financial Liability	Liability
ORA01	Prepayment Migration A/c	Prepayment Migration A/c	Control Accounts	Non Financial Assets	Assets
ORA01	Salaries and wages - Sharing	Salaries and wages - Sharing	Salaries and wages - Sharing	Employee Benefit Expenses	Expenses

Thanks a lot for your help!

[sturlaws]

what do the code for  [(Cumulative) Closing Balance] & [(Cumulative) Opening Balance] look like?

[Anonymous]

@sturlaws 

(Cumulative) Closing Balance =
CALCULATE([Closing Balance]|
filter ( ALL('Date'[Date])|'Date'[Date] <= max ('Date'[Date]))
)

(Cumulative) Opening Balance =
CALCULATE([Closing Balance]|
filter ( ALL('Date'[Date])|'Date'[Date] <= (max ('Date'[Date])-1))
)

where Closing Balance =

Var Amount = Sum(Accounting[Home_Currency_Amount]))

Var Debit_Amt = CALCULATE([Amount],Accounting[Debit_Credit]="DR")

Var Credit_Amt = CALCULATE([Amount],Accounting[Debit_Credit]="CR")

Return

Debit_Amt - Credit_Amt

[sturlaws]

hmm,

so the way the opening and closing balance is calculated, is to sum up all prior transactions. So if a report user is trying to see the average daily balance for a month, using averagex, this process is repeated for each day. And if the report is drilled down to GL Code Level 1, the number of calculations is multiplied with number of GL Code Level 1 visible in the visual. With a table with small number of rows it will work. With a table with a high number of rows, it will be slow and you might run out of memory.

I think you will have to create a table containing the opening and closing balance for each account for each day in order to imrpove the query speed.