Scree plot from proc corresp

[Nikster101]

Hi there. I want to run a multiple correspondence analysis. I cannot use proc corr as my variables are all categorical. I am using porc corresp but I am unsure of how to generate a scree plot. Below is my code

proc corresp mca data=clean2 dimens=2 outc=newout all;
tables race gender nsfasyn deptname campusname quintile;
run;

What procedure would be better suited for MCA and how do I get a scree plot?

[PaigeMiller · Accepted solution]

I think this is called the Inertia table (and plot) in SAS PROC CORRESP. Please take a look and see if that's what you are looking for.

Also, here's an example where it is calculated https://documentation.sas.com/doc/en/pgmsascdc/9.4_3.4/statug/statug_code_crspex1.htm

[Nikster101]

It is not quite what I am looking for. The Plot I need is a scree plot similar to that in principal component analysis with the 'elbow'. This was helpful though as I am now getting the inertia table

[ChrisHemedinger]

Check out @Rick_SAS post here - it might have some guideposts for you:

https://blogs.sas.com/content/iml/2017/08/02/retain-principal-components.html

[PaigeMiller]

---

@Nikster101 wrote:
It is not quite what I am looking for. The Plot I need is a scree plot similar to that in principal component analysis with the 'elbow'. This was helpful though as I am now getting the inertia table

---

Why is it "not quite" what you are looking for? It seems to me to be exactly what you are looking for.

[Nikster101]

Well I need a plot like this, 

i.e. A scree plot. I am not unsure of how to generate this plot in the proc corresp function.

[PaigeMiller]

That's exactly what they give you; except they use bars to represent the value instead of a series of lines to represent the value.

[Nikster101]

Yes I understand that, I was just wanting to know if getting an plot representation of it was possible.

[PaigeMiller]

So form over function? If it isn't a line plot, you can't use it? Even if the information is the same?

There is nothing stopping you from plotting it yourself using PROC SGPLOT and then it can look exactly the way you want it to look.