Skip to contents

Adjusted LS fitting of ellipsoids, by Kukush et al. 2003.

Usage

ellipsoid_ALS(x, s2, ..., just_U = FALSE)

Arguments

x

Point locations, n x d matrix

s2

Error variance, or a vector over which to optimize.

...

ignored

just_U

For internal use: Return just smallest eigenvector and bhat.

Details

Not really optimised yet. Good grid for s2 optimisation is something like seq(s2 * 0.1, s2, length=20) where s2 is the OLS estimate (access via $ols_fit$s2).

References

Kukush et al (2003): Consistent estimation in an implicit quadratic measurement error model, Computational Stat & Data Analysis