bigRfit: Rank-based estimation for linear models with big data.

In kloke/bigRfit: Big data version of Rfit

Rank-based estimation for linear models with big data.

Description

An implementation of rank-based estimation when working with big data. Uses step scores to increase the speed. Requires at least 2003 records in the dataset.

Usage

bigRfit(x, y, B = 1001, scores = Rfit::wscores, max.iter = 100, eps = (.Machine$double.eps)^0.625)

Arguments

x	n by p design matrix
y	n by 1 response vector
B	number of breaks (number of 'buckets' + 1)
scores	an object of class scores
max.iter	maximum number of iterations
eps	stopping criteria

Value

coefficients	estimated regression coefficents with intercept
residuals	the residuals, i.e. y-yhat

:

fitted.values	yhat = x betahat
scores	score function used in estimation
x	design matrix w/ intercept added (ie cbind(1,x))
y	original response vector
tauhat	estimated value of the scale parameter tau
taushat	estimated value of the scale parameter tau_s
symmetric	currently set to TRUE (needed for helper functions down the line)
iter	number of iterations
D1	final model dispersion
D0	null model dispersion
converage	convergance status (logical)
qrx1	result of call to qr of cbind(1,x)

Author(s)

John Kloke [e-mail]

References

Hettmansperger, T.P. and McKean J.W. (2011), Robust Nonparametric Statistical Methods, 2nd ed., New York: Chapman-Hall.

See Also

Rfit

Examples

n <- 10^4; p <- 10
x <- matrix(rnorm(n*p),ncol=p)
y <- rnorm(n)
#system.time(fit0 <- rfit(y~x))
#summary(fit0)
#system.time(fit1 <- bigRfit(x,y))
#summary(fit1)

kloke/bigRfit documentation built on April 20, 2023, 4:33 p.m.