Minimum Distance Least Squares Surface Fitting.

摘要

A least squares fitting technique is described where all of the experimental variables are considered subject to error, rather than the customary single variable. The procedure is particularly valuable when the actual physical measurements contain real and estimable errors in more than one variable. The fitting procedure minimizes the closest (perpendicular) distance from the experimental point to the locus on the assumed functional relationship among the n different variables. The locus is approximated by a (n-1) dimensional surface. The techique is adaptable to any number of variables and any number of adjustable constants in the assumed equation.