If I have $n$ variables $x_1,\dots,x_d$ in my dataset, and I regress the first one against all the others i.e.

$x_1=a_{12}x_2 + a_{13}x_3 + \dots + a_{1d} x_d + \epsilon_1$,

how can I calculate $\text{Var}(\epsilon_1)$?

If there are $n$ datapoints for each variable, then it is possible to calculate $\Sigma_{2:d}$ i.e. the covariance matrix for variables $x_2,x_3,\dots,x_d$. I can also easily calculate an unbiased estimate for $\text{Var}(x_1) = \frac{1}{n-1} \displaystyle\sum_{i=1}^n(x_1^i - \bar{x}_1)^2$.

But I don't know what to do with these?


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.