Recovering original matrix from PCA components

52 Views Asked by At

Suppose we have done the principal component analysis for data $X\in R^{n\times m}$ into mean vector $\bar{X}$, principal components $Q$ and diagonal matrix for eigenvalues $\Lambda$. My question is that how can we get back to original data $X$ by using those? Specifically saying, can we recover the original data by using $\bar{X}, Q$, and $\Lambda$?

Thank you in advance