Visual Filter for Cumulative Sum Issue

[Anonymous]

Hi,

I need some help with cumulative sum line.

Before selecting any slicer, the visual shows this:

But after applying filter, the visual shows the below.

How then can I make the line graph show the full graph, starting from zero to the end of the chart, similar to the first screenshot?

Regards,
Nicholas Hiew

[parry2k · Accepted solution]

@AnonymousUpdate cummulative total DAX as give below and that will do it, as @Seward12533 suggested to use cross filter to both direction, it has performance hit but instead of setting up the relationship, you can use it in DAX formula basically you are using cross filter to both direction when required

Cumulative = 
CALCULATE(
SUM('Table2'[Count]),
CROSSFILTER(Table1[Week of Year],Table2[Week No.], Both),
FILTER(ALL('Table2'),
'Table2'[Week No.]<=MAX('Table2'[Week No.]))
)

 

[Seward12533]

Also what is your desired result? do you want the **bleep** to be unaffected by the filter or do you want the scale to be fixed?  The problem is the **bleep** is being plotted on its own axis and PBI is dynamically adjusting it based on the data. Or do you want the other data to not be displayed and only show the filtered data?

If you want the **bleep** to adjust.

Look at either turning off the secondary Y axis to you **bleep** is plotted on the same scale (probably not what you want) or fixing the scales manually

If you want the **bleep** not be affected make sure you the USE the ALL filter(s) for the slicers your using

If you want the other data to be filtered out then adjust the visual interations and chose filter.

[parry2k]

Can you share sample data and what is your current measure?

[Anonymous]

Hi Both,

This is my desired result (lines in red), when sliced by category.

My Data (Table 1) looks like this, with the last column "Week No." is created in query editor:

And I have another table (Table 2) created to give me the following value (only the first column is from an excel sheet):

For cumulative column (note that I use excel to replicate these), I used the following measure:

Cumulative =
CALCULATE(
SUM('Table 2'[Count]),
FILTER(ALL('Table 2'),
'Table 2'[Week No.]<=MAX('Table 2'[Week No.]))
)

EDIT: I also tried adding a column for cumulative, which still yield the same visual results:

Cumulative =
CALCULATE(
SUM('Table 2'[Count]),
FILTER(ALL('Table 2'),
'Table 2'[Week No.]<=EARLIER('Table 2'[Week No.]))
)

Appreciate your help. Thanks.

Regards,
Nicholas

[Seward12533]

Rather than have the the week number come from your fact/data table have use the week number from the date table.

The **bleep** Measure would be following this pattern.

**bleep** = CALCULATE(sum(fact[value],all(date),filter(date,date[date]<=max(date[date]))

In your case since you only want cume YTD you would need to add a condition to filter out prior and future years &&date[year]=year(now())

So it would look something like
**bleep** = CALCULATE(sum(fact[value],all(date),filter(date,date[date]<=max(date[date]&&date[year]=year(now()))

The problem your having is your slicers are eliminating the data so your current measure for calculating the **bleep** will return null so nothing is displayed the measure based on the dates from the date table will return the results you require and you let the filter context of the PowerBI engine do the work an you don’t need to deal with the secondary table or the complexities of earlier. (Which is awesome that you figured out I still don’t have my head fully around that. )

[parry2k]

Hey @Anonymous Can you share the data in excel sheet for the solution?

[Anonymous]

Hey,

Here you go:

https://drive.google.com/file/d/1gr0UHXhg-Va_aLt-0QENIxqKVtKZtCLu/view?usp=sharing

Regards,
Nicholas

[v-piga-msft]

Hi @Anonymous,

It seems that you want to have a visual filter with cumulative Sum?

Could you share a screenshot of your visual filter setting?

Actually, if you have a visual filter for cumulative sum, it will show the values match the filter condition.

Best  Regards,

Cherry

[Anonymous]

Hi @v-piga-msft,

I used a donut chart based on the category. So for example, when no category is applied, it shows this:

When I choose a category, say for category A:

Now obviously this looks fine since this constitutes most of the values, however the cumulative line is still not right, as it still shows the total, not total by category.

This becomes ugly when I apply to Category C:

My slicer looks like this:

Hope my explanation is clear enough.

Regards,

Nicholas Hiew

[v-piga-msft]

Hi @Anonymous,

Do you want to get the output like below?

When I click "B" on the Donut chart, it will show this.

When I click "D" on the Donut chart, it will show like below.

If you want the output above, you could have a reference of the attachment.

Best  Regards,

Cherry

[Anonymous]

Hi @v-piga-msft,

I would like the cumulative line to be dynamic as well. Meaning when I choose category A, the cumulative line changes to Category A's cumulative line.

[Seward12533]

Hi, it has to do with the filter context not applying the way you think it will.  When you  use the slicer on one table it does not force that relationship backwards against the arrows.  There are may ways to handle it. The simlest is to edit the relationship so it crossfilters in both directions. This however is not the most robust solution.  Take a look at this tutorial I put together for an internal user group for some more robust solutions that will work as you develop more complicated models.

https://1drv.ms/u/s!AuCIkLeqFmlhhJgzLZ6jPcWgLGu3Yw

[parry2k · Accepted solution]

@AnonymousUpdate cummulative total DAX as give below and that will do it, as @Seward12533 suggested to use cross filter to both direction, it has performance hit but instead of setting up the relationship, you can use it in DAX formula basically you are using cross filter to both direction when required

Cumulative = 
CALCULATE(
SUM('Table2'[Count]),
CROSSFILTER(Table1[Week of Year],Table2[Week No.], Both),
FILTER(ALL('Table2'),
'Table2'[Week No.]<=MAX('Table2'[Week No.]))
)

 

[Anonymous]

Thanks @parry2k and @Seward12533!

BTW @parry2k, do you know how to make the line start from week 1 as well?

[Seward12533 · Accepted solution]

@Anonymous try this

Cumulative = VAR Result = 
CALCULATE(
SUM('Table2'[Count]),
CROSSFILTER(Table1[Week of Year],Table2[Week No.], Both),
FILTER(ALL('Table2'),
'Table2'[Week No.]<=MAX('Table2'[Week No.]))
)
RETURN
IF(Result,Result,0)

[Anonymous]

Awesome! Thanks @Seward12533!

[Seward12533]

Hi @v-piga-msft just to clarify I said the cross filter was the easiest. But my attachment compared several options and recommended forcing the context via DAX or if possible slice off a bridge table that forced a standard relationship down into each of the tables.   One question I have is the performance differecne of using CROSSFILTER vsreferencing the other table name (which I assume implies a cross filter)  This is the method I currently favor but if there is a performance benefit to using crossfilter vs just the table name I will start using that.

[Anonymous]

Here is the pbix file if anyone is interested:

https://drive.google.com/file/d/1pT9GoD_vD0DZfVpwDKHkzDwuuU63nnEB/view?usp=sharing