Simple Sankey plot help

[Maria Kristensen]

I want to create a Sankey diagram in Stata. I have installed the Sankey package and have been inspired by GitHub: GitHub - asjadnaqvi/stata-sankey: A Stata package for Sankey diagrams. I have a database with 497 patients, each having a unique ID from 1-497 to differentiate them. Each patient has a characteristic variable, "x," which is a count that can be counted at different tissue sections for each patient. When I count at tissue section 1 for all patients, I categorize the count (into 4 outcomes) in variable A. Similarly, for tissue section 2, I categorize the counts for each patient in variable B, and the same for tissue section 3 in variable C. Variables A, B, and C thus contain the same 4 categories, but counted at different tissue sections. With a Sankey diagram, I want to show how this categorization changes across the different sections while being able to track each patient. Therefore, I would like to create a Sankey diagram in Stata that displays the movements between variable A, variable B, and variable C. I have managed to achieve this using these commands:

sankey id, from(variable A) to(variable B)
sankey id, from(variable B) to(variable C)

The values that Stata assigns on the diagram don't seem to correspond to anything specific. I am not sure if it adds the ID numbers or something else. However, I can adjust this manually in PowerPoint afterwards, so it is the least of my concerns. I can combine the two diagrams manually to see movements from A to B and from B to C. But I want to visualize variable A all the way from A through B to C. Is there any way to do this based on my textual variable? I have tried a few different approaches, but I get errors.

I have attached the current plots corresponding to the stata code previous

Best regards, Maria

[Nick Cox]

Please post .png graphics attachments as explained in 12.4 within https://www.statalist.org/forums/help#stata

It seems to me that your categorization could be throwing away a great deal of the information unless there is a strong clinical or scientific rationale for it.