5 Replies Latest reply on May 24, 2016 10:59 AM by Tom W

# Design Structure Matrix

I've checked out multiple forums on here, and cannot find exactly what I'm looking for. Bora Beran's posts have proven to be helpful in some ways, but have not solved this dilemma. I've attached a simplified version of what I am trying too within Tableau.

The tabs work in a progressive order and ultimately, I am trying to replicate either the last or second to last tab. Formulas are set on manual and all equations were taken from a book called "data smart." Thanks for any help.

A further in depth explanation: I have a group and subgroup which are concatenated to form a unique field. "Instances" then are able to occur in multiple or no groups. That is transferred to create a matrix table which shows when an instance is shared between multiple coordinates. The concatenated group/subgroup then becomes both the X and Y axis and will show the similarity as to how often instances are shared between those two concatenated groups/subgroups.

• ###### 1. Re: Design Structure Matrix

What are you specifically having trouble with? How far have you gotten when trying to create this in Tableau yourself?

Right now you're asking a lot in terms of us reverse engineering your workbook to understand it so we can effectively build it for you in Tableau.

• ###### 2. Re: Design Structure Matrix

Hey Tom,

Sorry for the confusion. I was trying to add an example for playaround data. Design structure matrix - Wikipedia, the free encyclopedia - this link is a brief summary of what I am trying to do.

Thanks

• ###### 3. Re: Design Structure Matrix

I've attached an example of how I've replicated the example in Tableau and created the required Dataset in Excel to match.

This relies on a calculated field to determine the color (black or white) which I've called 'Color' and a calculated field called label to switch between the A,B,C labels for the black squares and the count of the dependency occurrences for the white squares. These both hand off a field called 'ElementMatchDependency' to match 'Element A' on the Y axis to 'A' on the X axis.

• ###### 4. Re: Design Structure Matrix

Is it possible to have the Element on both the X and Y axis and still populate the entire table? If you were to drag the Element pill into rows and columns, it only calculates the diagonal rows/columns; however, (Element A, Element E) both share a dependency and should have a number at that coordinate. Or is this not possible?

Thanks

• ###### 5. Re: Design Structure Matrix

You need some sort of a relationship between the element X and element Y. How would you propose to model that in your source data with only one dimension?