Covariance

A measure of how two variables vary together, Cov(X,Y)=E[XY]−E[X]E[Y]\mathrm{Cov}(X,Y)=E[XY]-E[X]E[Y]; its sign shows whether they tend to move in the same or opposite directions.

When to use

It appears in the OLS slope formula (the covariance of xx and yy over the variance of xx) and in the iid assumption that distinct observations have zero covariance. Zero covariance does not by itself imply independence.

See Lec 2.

Appears in

No references yet.