# Tag Info

Let $Y = X\beta +\epsilon$ be a multilinear regression model, where $Y = n\times 1$, $\beta=k\times 1$, $X = n\times k$ and $\epsilon$ is an error term. Then the least squares estimator, which minimizes the squared errors is $\hat\beta = (X^TX)^{-1}X^TY$. One can show that the maximum likelihood estimator is the same.