
1. Re: Tableau Control Chart with Multiple Dimension Values
Shinichiro Murakami Oct 1, 2018 1:40 PM (in response to Cameron Taylor)HI Cameron,
Without seeing data, it's quite difficult to assist.
Could you please attach mockup data as twbx and explain the expected value on case 2 with some real examples?
Thanks,
Shin

2. Re: Tableau Control Chart with Multiple Dimension Values
jason.harmer Oct 1, 2018 1:46 PM (in response to Cameron Taylor)While there might be a way to force the upper and lower control limit to not look at all dimensions (probably with a LOD calc), my question would be  do you really want all dimensions visible at the same time on that line graph? I think what's happening is that the dimensions is adding lines (one for each value of the dimension) on the control chart, which creates an UCL and LCL for each value. Would adding the dimension as a filter be a better execution of that?

3. Re: Tableau Control Chart with Multiple Dimension Values
Cameron Taylor Oct 1, 2018 2:13 PM (in response to Shinichiro Murakami)Attached  thank you! Basically, I would like one lower bound and one upper bound calculated for all values of the dimension. This would allow the values of the dimension to overlay one another on one line chart. In the real world example, the dimension has hundreds of values, and I would like to find the control points that are out of the ordinary for the whole population, not just one organization.

4. Re: Tableau Control Chart with Multiple Dimension Values
Cameron Taylor Oct 1, 2018 2:16 PM (in response to jason.harmer)Jason, thanks for the note. You are correct, I would like multiple lines for the dimension. I would like to find the control points for the population as a whole, but also still know which dimension value that control point belongs to. Essentially, I think I'm asking how to change the window function to account for the multiple values of the dimension. The average calculation likely needs to be updated to look at all dimensions in order to calculate a UCL and LCL.

5. Re: Tableau Control Chart with Multiple Dimension Values
jason.harmer Oct 1, 2018 2:27 PM (in response to Cameron Taylor)You might want to try a LOD to get the 'total' amount, and then reference this new amount in your upper and lower bound calculations.
Amount v2:
{ EXCLUDE [Organization]:SUM([Amount])}
Upper Bound: WINDOW_AVG(SUM([Amount v2])) + (WINDOW_STDEVP(SUM([Amount v2])) * 2)
Lower Bound: WINDOW_AVG(SUM([Amount v2]))  (WINDOW_STDEVP(SUM([Amount v2])) * 2)

6. Re: Tableau Control Chart with Multiple Dimension Values
Cameron Taylor Oct 1, 2018 2:34 PM (in response to jason.harmer)This results in something similar to the original. I think actually need to include the organizations somehow into the window average, as opposed to reporting on the total number. The total number is skewed by the outliers. I think the crux of the issue is...how do I create upper and lower bounds by both date and a dimension?

7. Re: Tableau Control Chart with Multiple Dimension Values
Shinichiro Murakami Oct 1, 2018 2:44 PM (in response to Cameron Taylor)If you simply want to show same upper / lower with Org as color, follow below simple removal of one measure.
==> sheet : Normal 2
If you want to change the upper/lower including the granularity of (week*Org)
Set table calc as below and do't forget synchronize the axis.
In this case, upper/lower changes.
==> Sheet : Desired with Flaws
Thanks,
Shin

Control Chart_SM_10.5.twbx 42.6 KB


8. Re: Tableau Control Chart with Multiple Dimension Values
jason.harmer Oct 1, 2018 2:54 PM (in response to Cameron Taylor)Hmm, maybe i'm not getting your problem statement. So, what do you want the UCL and LCL to represent? the total amount, regardless of organization? Or do you want it to represent something else (a subset of the organizations)?

9. Re: Tableau Control Chart with Multiple Dimension Values
Cameron Taylor Oct 1, 2018 3:39 PM (in response to Shinichiro Murakami)Thank you  the second one is what I'm after, I think! Just to be clear, could you please explain what the upper bound represents with the new table calculation? Is it the sum of amount divided by the number of weeks (52), and then divided again by the number of organizations (6)? This doesn't quite add up. For instance, if standard deviation is set to 0, the sum of amount (192163480) divided by 52*6 equals 615908. However, in the workbook, the bound shows 644844.
In my real world dataset, I have to set standard deviations to 20 in order to have any meaningful outliers, and I just want to make sure I understand what's going on.
Also, the upper and lower bounds are tied to Org 6...does this mean it's only using Org 6 numbers to calculate?
192163480 
10. Re: Tableau Control Chart with Multiple Dimension Values
Shinichiro Murakami Oct 1, 2018 3:33 PM (in response to Cameron Taylor)2nd one is based on
Date x Org combination as smallest data granularity
As reference, Another two examples of Lower ctrl value.
Across org in same date.
Across Week in same org (you will see multiple line as multiple org)
If you don't have Org in dimensions,(as original view)
distribution of Week total is plotted.
Thanks,
Shin

Control Chart_SM_10.5_v2.twbx 90.2 KB


11. Re: Tableau Control Chart with Multiple Dimension Values
Cameron Taylor Oct 1, 2018 3:49 PM (in response to Shinichiro Murakami)Thank you so much  please see the edit to my reply above. I seem to be off by a few thousand when calculating the total bound with standard deviation of 0. 644,844 instead of what seems to be correct which is 615,908. Is it because some weeks have less than 6 organizations? So it would really be at the granularity of 52 * 5.x?

12. Re: Tableau Control Chart with Multiple Dimension Values
Shinichiro Murakami Oct 1, 2018 4:02 PM (in response to Cameron Taylor)I m sorry I am not following, what do you mean by STV of "0" ?
If you want to calculate with very special formula, you can do trials and errors.
I am not quite sure what you really expect because I have no idea where your value (192163480, 615908, 644844) come from.
Thanks,
Shin

13. Re: Tableau Control Chart with Multiple Dimension Values
Cameron Taylor Oct 1, 2018 4:43 PM (in response to Shinichiro Murakami)Apologies...I meant standard deviation of 0. 192163480 is the total amount.
When I divide that by 52, I get the bound on sheet Normal (with SD of 0).
When I divide again by 6 (number of organizations), I get 614,908, which I
think should be the bounds with a SD of 0. However, the bounds with an SD
of 0 are showing as 644,844. Why is there a discrepancy?
On Mon, Oct 1, 2018, 7:11 PM Shinichiro Murakami <