How to Display top Reasons per category

I think we are missing important bits of your table. Is it indexed? Does it have a date? If it is Date, then does this correctly represent your question: We need the TopN rows per Category by Date, sorted by DeniedServices? (you can replace Date with index, if your table is indexed). 

In this answers, I assume you have your table indexed. If this is not the case, you can add an index column in the Query Editor of your table (Ribbon 'Add Column' -> Index Column). This gives us the following table:

Table Hospital: note the Index column!

Next we add an Parameter (and make sure it has 'Enabable load' activated, so it is loaded in the datamodel and we can reference it later on in DAX expressions):

Parameter with current value of 2.

Now we are going to make a calculated column in the Hospital table. This column will rank the index column per Category. The formula is;

RankPerCategory = RANKX(FILTER(Hospital, Hospital[Category]=EARLIER(Hospital[Category])), Hospital[Index], , ASC, Dense)

This results in the following table Hospital:

Note that the Ranking restarts for Category B rows.

Next step is to create a table derived from the Hospital table, one where we only want the TopN rows per category based on the Rank column. The N in TopN is based on the Parameter. The formula used to create the table and outcome (from the dataview):

FilteredHospitalTableRanked = FILTER(Hospital, Hospital[RankPerCategory] <= VALUES(TopNParam[TopNParam]))

EDIT: i forgot the part about the last sort. In the table FilteredHospitalTableRanked, we are going to do some cool stuff now; we are going to rank the rows based on two columns! First we are going to rank it based on the Category (so that A is always grouped an clustered before rows of B start) and then we are going to rank based on DeniedServices in that category. We are going to do this by creating yet another calculated column, this time in the FilteredHospitalTableRanked table:

Final Rank = 
RANKX (
    ALL ( FilteredHospitalTableRanked ),
    RANKX ( ALL ( FilteredHospitalTableRanked ), [Category],, DESC )
        + DIVIDE (
            RANKX ( ALL ( FilteredHospitalTableRanked ), [DeniedServices],, ASC ),
            ( COUNTROWS ( ALL ( FilteredHospitalTableRanked ) ) + 1 )
        ),
)

As you can see, we are using three times the RANKX formula. The first one is to create the overall ranking for each row. It ranks based on the (second RANKX) the rank of the category (A=1, B=2) + the normalized rank (meaning, from 0 to 1) of the DeniedServices in that category. The normalized rank is determined by ranking it normaly and then divide the rank by the total number of rows + 1. 

Putting it all in a table visual looks like this (sorted on 'Final Rank);

If this helped you, please mark this as the Solution 🙂 Kudo's are welcome!