iboss.od: The IBOSS method

In Ossifragus/IBOSS: Information-Based Optimal Subdata Selection

Description

This function implements the IBOSS method for the input covariate Z and response vector Y. It returns a list with elements: beta, the least squares estimate based on the subdata; se, the standard errors; sigma, variance estimate for the error term, index, index of the subdata.

Usage

iboss.od(Z, Y, k, int.adj = "TRUE")

Arguments

Z	the input covariate matrix or covariate vector
Y	the response vector
k	the subdata size
int.adj	whether to calculate the adjusted estimate of the intercept. It is TRUE by default.

Examples

library(IBOSS)
library(mvtnorm)
beta.true  <- rep(1, 51)
d <- length(beta.true) - 1
corr  <- 0.5
sigmax <- matrix(corr, d, d) + diag(1-corr, d)
n <- 5000
k <- 100
set.seed(0)
X  <- rmvt(n, sigmax, 2)
mu  <- beta.true[1] + c(X %*% beta.true[-1])
Y  <- mu + rnorm(n, 0, 3)
fit <- iboss.od(X, Y, k)
beta.od <- fit$beta
beta.odia <- fit$beta0.adj

Ossifragus/IBOSS documentation built on May 7, 2019, 9:28 p.m.