Multiple measure-based ‘Filters on this visual’ happening out of order

[Anonymous999]

My Data (table DT):

Location ID	Location Name	Date	WeekStarting	Week No.	Column ID	Open Units	Available Units
1	Location A	01-Jan-2024	31-Dec-2023	1	1	51	100
1	Location A	01-Jan-2024	31-Dec-2023	1	2	55	100
1	Location A	01-Feb-2024	28-Jan-2024	5	1	61	100
1	Location A	01-Feb-2024	28-Jan-2024	5	2	65	100
1	Location A	09-Feb-2024	04-Feb-2024	6	2	81	100
1	Location A	01-Mar-2024	25-Feb-2024	9	1	71	100
1	Location A	01-Mar-2024	25-Feb-2024	9	2	75	100
2	Location B	02-Feb-2024	28-Jan-2024	5	1	52	100
2	Location B	02-Feb-2024	28-Jan-2024	5	2	56	100
2	Location B	02-Mar-2024	25-Feb-2024	9	1	62	100
2	Location B	02-Mar-2024	25-Feb-2024	9	2	66	100
2	Location B	02-Apr-2024	31-Mar-2024	14	1	72	100
2	Location B	02-Apr-2024	31-Mar-2024	14	2	76	100
2	Location B	10-Apr-2024	07-Apr-2024	15	2	82	100
3	Location C	03-Mar-2024	03-Mar-2024	10	1	53	100
3	Location C	03-Mar-2024	03-Mar-2024	10	2	57	100
3	Location C	03-Apr-2024	31-Mar-2024	14	1	63	100
3	Location C	03-Apr-2024	31-Mar-2024	14	2	67	100
3	Location C	03-May-2024	28-Apr-2024	18	1	73	100
3	Location C	03-May-2024	28-Apr-2024	18	2	77	100
3	Location C	11-May-2024	05-May-2024	19	2	83	100

I also have the following measures:

SelectedWeekStart = SELECTEDVALUE(DT[WeekStarting])

LocationDateRank = RANKX( FILTER(ALLSELECTED(DT),

                                               DT[Location ID]=SELECTEDVALUE(DT[Location ID])),

                                               [SelectedWeekStart],,ASC,Dense )

DT% = SUM(DT[Open Units]) / SUM(DT[Available Units])

When looking at the rankings at the week-level without any filtering/slicing, everything looks good:

Next, when filtering on DT% >= 65% (via Filters on this Visual), the dynamic LocationDateRank is still correct:

But, when I also add LocationDateRank <= 2 to the ‘Filters on this Visual’ pane, the results are incorrect:

I want the result to be:

It seems the two measure-based filters are happening out of order (i.e., it is first applying LocationDateRank <=2 and then the DT%<= 65%). Changing the order of the filters listed in the ‘Filters on this visual’ pane has no effect.

How can I fix this so the LocationDateRank filter is applied last?

[Anonymous999]

In this case, the users will always expect the ranking to occur after all other filters (except the Rank filter itself) are applied. Once the filtered rows are ranked, they then want to only show the top X ranks per Location. I just don't know how to accomplish this.

[lbendlin]

What's the purpose of

LocationDateRank = RANKX( FILTER(ALLSELECTED(DT),

                                               DT[Location ID]=SELECTEDVALUE(DT[Location ID])),

                                               [SelectedWeekStart],,ASC,Dense )

Are you trying to find out the ROWNUMBER?

Both DT% and LocationDateRank seem to be immutable, and can thus be created as calculated columns.

[Anonymous999]

Thanks for the reply Ibendlin.

LocationDateRank is essentially the row number but is dynamic based upon the filters that have been applied. You'll notice that in the second screenshot (i.e., the one that shows the DT%>65% result), the rank (or row number) has changed from the 1st screenshot. Location A's two rows in the 2nd screenshot are numbered 1 and 2, but they were originally rows 3 and 4 in the first screenshot. 

DT% will also vary depending on aggregation and filters. The original column data includes a Column ID, which is the lowest level that downtime % is calculated. But when aggregating irrespective of Column ID or the WeekStarting date, the DT% can be different.

Hope this helps to clarify.

[lbendlin]

so you have two moving targets .  That will likely result in ranking changes.  You would need to define your expected result and then materialize the measures accordingly.