perccal_interval: Calculates Percentile-Calibrated Linear Regression Confidence...

In perccal: Implementing Double Bootstrap Linear Regression Confidence Intervals Using the 'perc-cal' Method

Description

This is the main function of the package. It takes as inputs the predictor/response matrix appended together, which can be either a data frame or a matrix, along with the desired coverage and other settings, and outputs marginal confidence intervals for each of the predictors, including the intercept.

Usage

perccal_interval(Xy, alpha, G = 20, B = 999, B2 = 999)

Arguments

Xy	[n by (p+1)] matrix: X in columns 1 to p, y in column p+1. X is the design matrix, and is assumed to not include a vector of one's.
alpha	Target coverage desired.
G	Number of grid points to evaluate calibrated percentile method on each side over.
B	Number of 1st stage bootstrap samples.
B2	Number of 2nd stage double bootstrap samples.

Value

Return a (p+1)x2 matrix containing confidence intervals for all regression coefficients, estimated via the perc-cal method.

Examples

set.seed(1234)
n = 32
B = 500
B2 = 500
G=20
x1=rnorm(n)
x2=rnorm(n)
eps=rnorm(n)
y = x1 + 2*x2 + eps
Xy = cbind(x1,x2,y)
alpha = .025
perccal_interval(Xy, alpha, G, B, B2)

perccal documentation built on May 2, 2019, 1:06 p.m.