Top N of sum of column per name per date

[Anonymous]

Let's say I have three columns:

NAME: Name of personel

SALES: Amount to be summed

DATE: date

I can not manage to make a top 10 of NAMEs that have the highest SALES per DATE

For example if this is data:

Name	Date	Sales
A	1	60
A	1	150
B	3	225
C	2	150
D	5	125
A	2	100

A top 3 should look like this:

Name	Date	Sales
B	3	225
A	1	210
C	2	150

I can make a table sorted by Sales of above three columns showing all data with correct sum of sales per date per name, however once I filter that visual by Name -> Top N - > top 10 by Sales, I get complete gibberish. How do I go about this?

[Anonymous]

OK so this is weird. Your solution DOES work, but not yet in the way I want it to. 

The top 10 is made from data from table A.

Table A contains a date column, which contains weirdly formatted dates that I don''t use anywhere in the report.

Table A is linked to column B through a Transaction ID (each transaction ID = one row that also contains name and sales amount). Column B has the correct date format for each transaction ID. The date format in column B is linked to a Date Table (dates, weeks, months, years) that I use throughout the report and seems to work correctly when summarizing data from Table A.

However... when I use the date column from the Date Table in my Top N table, I get gibberish values. If I use the date column from Table A, I get a correct top 10 table. Even though the date translation (date, weeks, months, etc.) seems to work correctly for all other visualizations that use data from Table A?

How do I make the table with the correct dates from the Date Table?

EDIT: I'll mark this as solved as a solution was given for the original question. I'll post the new question in a new topic.

[amitchandak]

@Anonymous , Can you share a bigger sample. I tried the measure I shared above and got the results

[amitchandak · Accepted solution]

@Anonymous , Try a new measure like this and filter for <=3 

rankx(allselected(Table[name], table[date]), calculate(sum(table[sales])))

For Rank Refer these links
https://radacad.com/how-to-use-rankx-in-dax-part-2-of-3-calculated-measures
ttps://community.powerbi.com/t5/Community-Blog/Dynamic-TopN-made-easy-with-What-If-Parameter/ba-p/367415

[Anonymous]

Thank you for the quick reply. I've seen this posted and tried it, but this doesn't give the desired result.

If I just make the table (no top N, no fancy filtering), the top 10 of sales per date per employee (in a very very large table) show the correct numbers. There are no duplicates of values in the top 10 of that very large table. However once I try your solution to make a actual top 10 list, the highest value is not on top anymore. There is a whole bunch of values that have ranking "1", i.e. I end up with a much larger list than a Top 10, even though there are no duplicate values in the actual top 10 ?