Hi Team,
Is there a way to perform PCA in JMP if the data format is matrix type? (Correlation matrix as input)
Thanks for your help
Murali
If mat is your matrix, then SVD(mat)[1] gives you the eigenvectors and SVD(mat)[2] the eigenvalues.
It might also be useful to know that the Structural Equation Modelling platform in JMP Pro can use a correlation matrix as the input. And you would be able to do something similar to PCA there, I think.
If you use @DonMcCormack 's suggestion, remember that JMP centers and scales the variables by default. You must perform this normalization on the matrix prior to the SVD.
My quick and direct answer is that JMP doesn’t take correlation matrix as an input to PCA. As for SEM it does not do PCA.
