R has a DModX Calculation for PCA. I've seen references in some online texts that a critical DModX can be used to flag samples which are significantly poor fit to the model. The R documentation states that DModX is approximately F-distributed. How would I use the F Distribution to calculate the critical DModX value at a particular confidence interval given my PCA model? It seems there are several definitions of DModX in different software packages, each using different scaling parameters. Sticking to the definition used in R would be appreciated.

  • $\begingroup$ If by "definition used in R" you mean the link you provided, then be especially careful, because--at least at first glance--the statistic it defines looks like it is the square root of an F-distributed variable. $\endgroup$ – whuber May 7 '19 at 17:09
  • $\begingroup$ One source did say that DModX² is proportional to F. This is exactly why I'm asking, Thanks for your insight. $\endgroup$ – Kevin Nowaczyk May 7 '19 at 18:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.