Pivot date problem

[gustavo80]

Hi everyone! I'm pretty sure that my problem could be solved through pivot or unpivot feature, but I encountered a problem that I wasn't able to fix.

My dataset looks like:

	installs	signups	players	sesions
30/5/2020	10	20	30	40
29/5/2020	20	33	44	22
28/5/2020	20	11	33	5
27/5/2020	15	10	22	6
26/5/2020	16	6	8	7

I need to show it transposed, like:

	30/5/2020	29/5/2020	28/5/2020	27/5/2020
installs	10	20	20	15
signups	20	33	..	.
players	…	..	.	.
sesions	..	..	..	6

Using pivot feature, seems to be ok:

But the problem, as the date columns now (due to pivot action) are metrics, I need to select one by one to include it in the table. But as my query datasource every day will return new days ( it recover last 4 days metrcs ) , the "new date" included every day, appears not checked ( as you can see 30/05/2020), so doesn't appear in the table.

Is there any solution to do it ? I mean, I need to show always the days returned into my datasource, with no manually intervention.

My .pbix attached, with the original table ( a sample ) and the pivoted table as other

source: https://drive.google.com/file/d/19VX0qKjk25MgowRCIxjBf6MWZ9na_v6Z/view?usp=sharing

Thanks in advance,

Regards!

[MFelix · Accepted solution]

Hi @gustavo80 ,

You must use the Unpivot column in the query editor.

• Add an index column
• Select the columns Installs, signups, players and session.
• Unpivot
• Get 4 Columns
  • Date
  • Index
  • Attribute
  • Value
• Add a custom column:

if[Attribute] = "installs" then 1 else
if[Attribute] = "signups" then 2 else
if[Attribute] = "players" then 3 else
4

• Sort the Date by the last index column and the attribute by the previous column you have created
• Make you matrix with the following setup:
  • Rows: attribute
  • Columns: Date
  • Values: Values

Check PBIX file attach.