erfe {erfe}

Dexpectilize a vector according the a single asymmetric point

Description

This function is the main fucntion of the erfe package. It estimates the ERFE model for a panel dataset and for a sequence of asymmetric point \tau \in (0, 1). When \tau=0.5 the function estimate the classical within-transformation estimator and its sandwich covariance matrix.

Usage

erfe(predictors, response, asymp = c(0.25, 0.5, 0.75), id)

Arguments

predictors

Numeric matrix of covariates/regressors.

response

Numeric vector of response variable.

asymp

Sequence of asymmetric points.

id

Ordered vector of subject ids.

Value

Returns a list of list according to the asymmetric points. Each list has objects related to the erfe model such as the asymmetric point, the coefficient-estimate, the standard deviation, the estimated covariance.

Author(s)

Amadou Barry, [e-mail]

References

Barry, Amadou, Oualkacha, Karim, and Charpentier Arthur. (2022). Weighted asymmetric least squares regression with fixed-effects. arXiv preprint arXiv:2108.04737

Examples

set.seed(13)
temps_obs <- 5
n_subj <- 50
sig <- diag(rep(1,temps_obs))
id <- rep(1:n_subj, each=temps_obs)
rvec <- c(mvtnorm::rmvnorm(n_subj, sigma = sig))
fvec <- (1 + rep(rnorm(n_subj) , each=temps_obs))
predictors <- cbind(rt(n_subj * temps_obs, df=2, ncp=1.3),
 1.2 * fvec + rnorm(n_subj * temps_obs, mean = 0.85, sd = 1.5) )
response <- 0.6 * predictors[, 1] + predictors[, 2] + fvec + rvec
asymp <- c(0.25,0.5,0.75)
erfe(predictors, response, asymp=c(0.25,0.5,0.75), id)