next up previous
Next: Multivariate phd Up: Methods available Previous: Principal Hessian direction

Quadratic phd

Li (1992) proposed an alternative method of estimating the Hessian matrix based on quadratic regression, as follows: (1) fit the ols regression of the response on the full second-order model based on all $p$ predictors; (2) set $\hat{M}$ to be the $p\times p$ matrix whose $(i,j)$ element is the estimated coefficient of $x_ix_j$. Given this estimate of $M$, proceed as with other phd methods.

Sandy Weisberg 2002-01-10