Matrix Power BI totals are not calculated - matrix

Question

the sub-totals and grand-totals of the CA facturé, of the provision sur CA return 0 for the Dimfi project code which contains several lines.
On the other hand, for the Dimfi project code which contains a single line, the subtotal is displayed correctly.
Here is a formula I used
CA Facturé =
IF(
OR(
CONTAINSSTRING(SELECTEDVALUE('ECRITURE COMPTABLE'[N° de document]),"AV")
,CONTAINSSTRING(SELECTEDVALUE('ECRITURE COMPTABLE'[N° de document]),"VT")
)
,[Solde]
,0
)
enter image description here
enter image description here
when i replaced SELECTEDVALUE() by MAX() it gaves this result, the total of CA FACTURE is incorrect

Answer 1

Some ideas before for measure commenting.
DAX calculates result separately for each row you have in the matrix.
On a screenshot you have 23 rows, so the measure calculates 23 times. Each row has it own context (some values that affect the result).
When DAX calculates row - 1, the context of calculation is Dimfi code project = E002P00019-01
So DAX filters your table before measure returns you a value like this:
CALCUATE(
[CA Facturé] -- the body of the measure works after filtering
,'ECRITURE COMPTABLE'[Dimfi code project] = "E002P00019-01"
-- the table consist of unique [Dimfi code project] value
-- and multiple dependent [N° de document] values that are hidden in the matrix.
)
When DAX calculates row - 6, the context of the calculation is Dimfi code project = E002P00021-01 and [N° de document]=AVPREE0020001795
So DAX filters your table before measure returns you a value like this:
CALCUATE(
[CA Facturé]
,'ECRITURE COMPTABLE'[Dimfi code project] = "E002P00021-01"
,'ECRITURE COMPTABLE'[N° de document]="AVPREE0020001795"
-- the table consist of unique [Dimfi code project] value
-- and unique [N° de document] value
-- may be you get a table with 1 row
)
For Total you have all table for the measure. The table is not filtered.
Now comments:
CA Facturé =
I want to collect all visible in the context unique values for [N° de document] with VALUES() function.
It's about the same as DISTINCT(),
but DISTINCT() excludes blank values from the result.
VALUES() will return all unique values (codes) for total row in the matrix,
multiple [N° de document] for rolled up [Dimfi code project],
1 value if you see [N° de document] in the matrix row:
VAR docNumValues = VALUES('ECRITURE COMPTABLE'[N° de document])
docNumValues is a table with 1 column and all unique [N° de document] that are visible in the context
for row 5 of the matrixit it looks like this:
N° de document
AVPREE0020001795
REPOE0020001164
VTPRE0020009986
...
VTPRE0020018283
ADDCOLUMNS in this measure creates a virtual table.
It consist of as that much row as you have unique [N° de document] values in a certain row of matrix.
ADDCOLUMNS is an iterative function it means that goes through each row
of base table docNumValues and evaluate an expression that is presented after new column name.
the first iteration of ADDCOLUMNS will add a filter to CALCULATE like this:
CALCULATE(
IF(...
)
,`[N° de document]=AVPREE0020001795`
)
and CALCULATE will cut a table for calculation with a 1 unique value.
VAR withSum =
ADDCOLUMNS(
docNumValues
,"#Sum",CALCULATE( -- adds filter `[N° de document]=...
IF(
OR(
LEFT(
SELECTEDVALUE('ECRITURE COMPTABLE'[N° de document])
,2
)="AV"
,LEFT(
SELECTEDVALUE('ECRITURE COMPTABLE'[N° de document])
,2
)="VT"
)
,[Solde]
,0
)
)
)
AS a result of the ADDCOLUMNS() for row 5 of the matrix you will have a virtual table withSum like this:
N° de document
#Sum
AVPREE0020001795
-5823.10
REPOE0020001164
0
VTPRE0020009986
10 493.55
...
...
VTPRE0020018283
5505.07
Than you just make a summation of all values that you have in the table.
RETURN
SUMX(withSum,[#Sum])
I used LEFT() instead of CONTAINSSTRING(), because it's safety - it can be that you will have the code like RTVT, and CONTAINSSTRING() will return TRUE()