Formula to sort by accuracy (absolute error)

[DIFF1]

Hello everyone, 

I need some help.  Any advice will be appreciated, especially if someone can explain. 

I have such a visualization. 

I need to sort by lowest forecast accuracy. By saying this, I need to somehow take the forecast accuracy of 100% as the best one and to rank the ratio of Accuracy considering that items can be over and under forecasted. So 100.20 and 99.80 basically show that there is an error of 0.20% and it should be considered similar level issue. 

Measure of Accuracy is 1-(FCST-ACTUALS)/FCST 

Does someone know how i can sort it taking into consideration this absolute error...

[DIFF1]

@ebeery you know. It works. Thank you very much!

[DIFF1]

@ebeery  Thank you very much for your reply. 

I am not sure but something is totally wrong, it doesn't calculate correctly. That's what I have now. 

Calculation for Accuracy

And calculation for Error

[ebeery]

@DIFF1 What do you mean by "it doesn't calculate correctly"?  What is the actual result you are expecting?  It seems to me like it is operating as intended - the are sorted by worst to best accuracy.  To see the most accurate values just sort in ascending order.

[DIFF1]

@ebeery  Thank you very much. But the first step is how to basically let PowerBI know that 100.01 accuracy and 99.99 accuracy is basically a mistake of 0.01 so that it is treated as the same mistake? That is the biggest issue for me. I don't understand how it can be done.

[ebeery · Accepted solution]

@DIFF1 Why not something using absolute value? Like:

Error = ABS([Accuracy] - 1)

[ebeery]

@DIFF1 assuming that you don't want the absolute error to be shown in the visual, the only way I'm aware of with a table/matrix visual is described in the 2nd part of the article below.   Basically involves creating a measure for sorting, adding to the visual, and adjusting the column width such that the column is no longer visible.

https://pbivisuals.com/2021/03/03/power-bi-sorting-data-by-a-column-not-shown-in-visual/