Pivot table and create co-occurencies heatmap

[Anonymous]

I have the following table

RECORD_ID	TAG
1	TAG1
1	TAG2
2	TAG1
2	TAG3

I would like to get a heatmap of co-occurrencies. In matrix form, it would be

	TAG1	TAG2	TAG3
TAG1	count of record_id with tag1	count of record_id with tag1 and tag2	count of record_id with tag1 and tag3
TAG2	count of record_id with tag2 and tag1	count of record_id with tag2	count of record_id with tag2 and tag3
TAG3	count of record_id with tag3 and tag1	count of record_id with tag3 and tag2	count of record_id with tag3

which, in this case, is a symmetric matrix. Then I would like to color the cell based on the value.

I guess to obtain that matrix I should pivot my table on TAG and use those counts as values, but I am not familiar with PowerBi and DAX, so how can I achieve this? and how can I get the coloured heatmap? Is there a more direct way to achieve my goal?

[Greg_Deckler · Accepted solution]

@Anonymous - See PBIX attached below sig. Tables (31) and (31a). First, create disconnected table from original like this:

Table (31a) = DISTINCT('Table (31)'[TAG])

Then this measure:

Measure 31 = 
    VAR __Table1 = SELECTCOLUMNS('Table (31)',"Record",[RECORD_ID])
    VAR __Table2 = SELECTCOLUMNS(FILTER(ALL('Table (31)'),[TAG]=MAX('Table (31a)'[TAG])),"Record",[RECORD_ID])
RETURN
    COUNTROWS(INTERSECT(__Table1,__Table2))

Put [TAG] from (31) in Rows of a Matrix, [TAG] from (31a) in Columns, Measure 31 in Values.

[Anonymous]

Thanks @Greg_Deckler! may I ask you to elaborate a bit more on what the variable __table2 is? I am not quite sure what happens there, and I'd like to understand the process. Thank you.