While reading about some tests like Bartletts test of sphericity to determine whether to apply PCA or not,I came to know that the test cannot be used if some conditions are not met.
I ran this test on a dataset in which there are highly correlated variables and this is the output:
Since the p-value is less than 0.05 we can do PCA as the correlation matrix is significantly different from the identity matrix,but I am afraid the n:p condition is not met in this case as there are only 15 columns and 2600 rows.
What similar tests which compare the correlation matrix with the identity matrix can be used here??
Can someone please guide me on this??