MATH 423 Lecture Notes - Lecture 21: Covariance
Document Summary
Hy. is the hat matrix or the influence matrix. Thus hij isthe at whichthe ith fitted value changes as we vary the in observation the influence thatobservation has on that. Ted value f oiin z symmetry r sirlh. is. Asymmetric idempotent matrix is a projection matrix this means that h projects y into a lower dimensionalsubspace i hy is a linearcombination. Specifically y is a point in rn but of no vectors namely the two columns of x in otherwords. H projects y onto the column space ofx. The column space of x is the setof vectors that can bewritten as linear combinations of x. The vector of residuals is eiy y y hy. Hereare some propertiesof eh i influence i hj li hitei h z symmetry. Y y hy eh eh lehllxfteecji. Eili hixftli hiej xp xxixixtxfoer r i cancel. eu. Thusthe variance of each residual is notquite g nor are the residuals exactly uncorrelated htt. Lie i h c.