building decomposition trees using randomly generated input

[박팀]

Hi,

I'm a teacher who wants to help students improve their pronunciation skills. One of the ways we can do this is by using something called 'minimal pairs'. These can be used to create a 'minimal pair word tree' (aka a decomposition tree in Power BI), or something that looks like this:

https://drive.google.com/file/d/14zoANAJPT13S_zg-6PtgZNX8doBxMGH8/view?usp=sharing

I have created a dataset (100,000 rows in excel over some 500 pairings):

https://docs.google.com/spreadsheets/d/14F7yHA52g64Qbck6vDpKQeQvOS7vy0n0/edit?usp=sharing&ouid=10280...

I would like to be able to create a tree using say, 16 random pairings in any given set. How easy/hard is this going to be?

Please also find a Power BI file that I've been working on (at the moment this is essentially the same as the Excel file above):

https://drive.google.com/file/d/1T1AXoWC3dStvLh_kigEfWgxXmi1pM9SJ/view?usp=sharing 

[lbendlin]

You can create random numbers in DAX that get updated with each dataset refresh.

I would like to be able to create a tree using say, 16 random pairings in any given set.

Your example looks more like four pairs?  Do the individual hierarchy layers need to be phonetically related?

[박팀]

Thank you for your response.

The example I gave wasn't entirely accurate; any tree that is drawn should select (randomly) from any given set, a number of pairs as needed. For example:

1 layer = 2 pairs

2 layers =  pairs

3 layers = 8 pairs

4 layers = 16 pairs

5 layers = 32 pairs

beyond 6 layers, it may be difficult to fit everything on one page, or at a size that is legible. 

A basic workflow would be as follows:

1) user selects 2 phonemes to compare, say, 'L' and 'R'. They may also like to select the number of layers/pairs

2) a tree is drawn (as shown in the example PDF). No one pair should be used more than once.

3) it would be nice to be able to 'refresh' or 'redraw' the tree as needed by clicking a button.

[lbendlin]

Here is what I came up with. I am pretty sure this is not what you intended, but that's how decomposition trees work in Power BI. 

The random selections will work on the Power BI Desktop when you click "Refresh".

[박팀]

Thank you very much for the screenshot and attachment.

You're right - it's now quite what I'm after so it might be back to the drawing baord. An upshot of this is I've discovered Power BI (and how useful it is) and will be using it for some other ideas in the near future regarding data visualization. It might be a little ambitious but I'd like to try drawing some of the items discussed here:

https://datascience.stackexchange.com/questions/38264/visualizing-many-one-to-many-relationships

[lbendlin]

I would certainly go with network graph visuals.