I am attempting to verify a calculation found at another question on this site. The formula is said to provide a correlation matrix using
q = D-1ED-1
where q is the correlation matrix;
E is the covariance matrix obtained from ATA (where A is the original data matrix and AT is the transpose of A. (The source uses a capital sigma instead of E, but I don't know how to insert that character here)
D is the square root of each diagonal value in E, and D-1 is the inverse of D. D-1 must be a square matrix with values in the diagonal, but all others set to zero.
Beginning with my original data, I used Excel to create correlation matrix #1. Then, using this method and Excel, I calculated correlation matrix #2. The two are not the same. The Excel calculations use MMULT(Array1, Array2) for multiplication, and MINVERSE(Array) to obtain an inverse. I have checked these two functions and they work properly. There are too many steps in the Excel sheet to try to include them here.
Checking around the internet I have not yet found as clear description as that at the link on this site, and yet I am unable to verify the calculations.
I must be missing something on how q is found, and look forward to suggestions on how to do this.
The other question I referenced is at Correlation matrix from Covariance matrix