Principal Component Analysis in R- Data rotation

[inferno]

I ran prcomp on my data (7000 observations, 48 variables), and I biplot I got seems to have thousands of points. My guess is that R is treating each of the 7000 observations as variables. Is there a way to "rotate" my data so that I get 48 points on my PCA graph instead of 7000?

Each column in my spreadsheet represents a variable, and each of the 7000 rows represents a specific gene.

 

[Lazar]

https://tgmstat.wordpress.com/2013/11/28/computing-and-visualizing-pca-in-r/

 

[inferno]

IF there are 7000 obs and 48 variables, how many points should you expect to see on the PCA plot? 48, right?

 

[Lazar]

Depends on what I wanted to graph right? How many components did you extract? If you have more than 2 components graphing is not really going to work

 

[inferno]

I extracted 2 components.

 

[Lazar]

So the link I gave you give what you want right? I mean the last graph looks exactly like you want. It is a pretty good tutorial of the whole process.


In the future I would suggest not just asking questions but also giving the R code you are trying to use. This helps us help you. Put the code in code tags (click on the # sign). Also please give us a toy dataset to work with. See https://stackoverflow.com/help/mcve on how to provide a minimal reproducible example.

 

[inferno]

I read the link, I followed the tutorial. I simply want to confirm that my graph is correct. Hence I ask, if there are 7000 obs and 48 variables, how many points should you expect to see on the PCA plot? 48, or 7000?

I keep what you said in mind for future questions.

 

[Lazar]

Again depends. I would most likely want to see a graph like the last one in the link in which there are only the 48 data points. It might also be nice to see one with BOTH the 48 variables overlaid upon the actual cases i.e., 7000 much like the 2 graph in that link.