I want to make a covariance matrix that incorporates my belief of how correlated the various dimensions are. The reason why I want to incorporate my belief is that in my modelling, the dimensions are food items and I want to manually correlate which food items are similar to which other in my covariance matrix. Is there any way to generate a valid covariance matrix(ensuring positive definiteness, though symmetry is easy)? After making such a covariance matrix, I intend to do a Cholesky decomposition and use it to uncorrelate my data-points. Are there any ways to do an approximate Cholesky decomposition, given a non-positive definite symmetric matrix?

Any help would be appreciated!!!


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.