Hi
I would like to see whether my data is suitable for principal commpent analysis or not. For this purpose, I am using following three criteria: Bartlett's test, KMO (Kaiser-Meyer-Olkin) and Determinnant. KMO and
Determinnant tests are OK.
But when I run Bartlett's test using following R command I get warning message.
cortest.bartlett(df), where df is a data frame
The warning messages are:
$chisq [1]
NaN
$p.value [1] NaN
In log(detR) : NaNs produced
My undrestating is that when R is trying to take a log of determinant of correlation matrix of a data frame, when it encounters negative values then this type of warning messages are produced.
Could any suggest how to fix this issue please.
Regards
Abdur