.BG
.VE $Header: /usr3/s/current/s/.help/RCS/rbiwt,v 1.2 83/12/30 19:33:11 rab Exp $
.FN rbiwt
.TL
rbiwt: Robust Simple Regression by Biweight
.CS
rbiwt(x, y)
rbiwt(x, y, start, k, tol, iter)
.PP
.AG x
vector of observations on independent variable.
.AG y
vector of observations on dependent variable.
.AG start
vector giving starting values of intercept and slope.
Default, use least-squares `start'.
.AG k
biweight scale parameter, default 6.
.AG tol
convergence criterion. Default .001
.AG iter
maximum number of iterations. Default 20.
.RT
structure containing components `coef', `resid', and `wt'.
.RC coef
vector giving intercept and slope.
.RC resid
vector like `y' giving residuals from fit.
.RC wt
vector giving weights used in final weighted least-squares
step.
.PP
Does not fit data to more than 1 independent variable (see `rreg').
.SH REFERENCE
Coleman, D., Holland, P., Kaden, N., Klema, V., and Peters, S. C.,
"A System of Subroutines for Iteratively Re-Weighted Least-Squares Computations", 
.ul
ACM Trans. Math. Soft.,
Vol. 6, 327-336, 1980.
.SH SEE ALSO
Function `rreg' generalizes `rbiwt' to multiple regression.
.EX
plot(x,y)
abline(rbiwt(x,y))     #add line to plot
.KW array*
.KW regression
.KW robust
.WR
