topic Multi-Step Calculation in DAX Commands and Tips

Multi-Step Calculation

djbuncle — Wed, 21 Aug 2019 11:52:23 GMT

Hi. The issue I'm faced with is coming up with a calculation which uses 3 other calculations that are sliced on [Months Since Reg] (on columns) however the final result will be a single value.

To explain further;

Pivot [1] is my base [% Split] calculation respective to the [Months Since Reg]

Pivot [2] shows the [Value Per Player] over the [Months Since Reg]

Pivot [3] shows the multiplcation on [% Split] * [Value Per Player] with respct to the [Months Since Reg] to get [Adj Value Per Player]

Then finally I want to calculate my [Expected Value] in green which is the sum of all [Adj Value Per Player]

So, as you can see it's a multi-step process where I'd need to utilise several table variables to reach my result, however my DAX skills don't extend to such expertise.

Can anyone shed some light on how I may achieve this?

Thanks.

Re: Multi-Step Calculation

Stachu — Wed, 21 Aug 2019 14:45:12 GMT

Can you add sample tables (in format that can be copied to PowerBI) from your model with anonymised data? Like this (just copy and paste into the post window).

Column1	Column2
A	1
B	2.5

probably it will be something like this

Measure =
VAR __Summary =
    SUMMARIZECOLUMNS (
        'Table',
        'Table'[Country],
        'Table'[Months Since Reg],
        "Split", [% Split],
        "VPP", [Value Per Player]
    )
RETURN
    SUMX ( __Summary, [Split] * [VPP] )

Re: Multi-Step Calculation

djbuncle — Wed, 28 Aug 2019 09:29:25 GMT

Hey Stachu, thanks for your response - I've only got around to giving it a go now.

The added complexity here is that the 2 values to multiply are actual calculations themselves using other tables so I can't really supply some example data easily, but here's what the calculated values would be.

Country	MonthsSinceReg	Split	ValuePerPlayer
Germany	0	1.00	145
Germany	1	0.14	317
Germany	2	0.09	303
Germany	3	0.07	248
Germany	4	0.06	265
Germany	5	0.05	239
Germany	6	0.04	274
Germany	7	0.04	242
Germany	8	0.04	227
Germany	9	0.03	203
Germany	10	0.03	202
Germany	11	0.03	262
Germany	12	0.02	404
Germany	13	0.02	363
Germany	14	0.02	352
Germany	15	0.02	248
Germany	16	0.02	219
Germany	17	0.01	364
Germany	18	0.01	467
Germany	19	0.01	1,038

I ended up using SUMMARIZE instead of SUMMARIZECOLUMNS to get the calculation to work, but now when I view this in Excel it only seems to do the calculation in month 0. Note that when I'm pivoting in Excel I don't want to use the [Months Since Reg] attribute, this is just needed in memory to do the required calculation -> the end result is one value for Germany.

Thanks

Re: Multi-Step Calculation

Anonymous — Wed, 28 Aug 2019 11:38:52 GMT

Stachu, you should never use SUMMARIZECOLUMNS in a measure. The reason being that SUMMARIZECOLUMNS does not honor filtering context introduced by CALCULATE. This function is purely for QUERIES.

Best
Darek

Re: Multi-Step Calculation

Anonymous — Wed, 28 Aug 2019 11:43:27 GMT

djbuncle, please do not use SUMMARIZE for calculating with measures. You should follow the ADDCOLUMNS/SUMMARIZE pattern as outlined here:

https://www.sqlbi.com/articles/best-practices-using-summarize-and-addcolumns/

Best
D.

Re: Multi-Step Calculation

Anonymous — Wed, 28 Aug 2019 11:51:21 GMT

[Expected Value] =
	SUMX(
		ALL ( T[Months Since Reg ] ),
		[% Split] * [Value Per Player]
	)

You can try the above... but I think you'll need some adjustments with respect to other filters. I don't have enough information to tell you exactly what the formula should look like.

By the way, [Months Since Reg] must be an attribute in a dimension (not in a fact table).

Best

Re: Multi-Step Calculation

djbuncle — Wed, 28 Aug 2019 12:04:40 GMT

That did the trick Anonymous . Much appreciated.

Re: Multi-Step Calculation

Stachu — Wed, 28 Aug 2019 17:57:03 GMT

Anonymous
fair point