gapstat {daewr} | R Documentation |
This function computes the gap statistic which is used to test for an outlier using Daniels method
gapstat(beta, pse)
beta |
input - vector of coefficients from saturated model fit to the data |
pse |
input - Lenth's PSE statistic calculated from the elements of beta |
returned gap statistic
John Lawson
## The function is currently defined as function (beta, pse) { p <- length(beta) psehe <- pse sel <- beta >= 0 betap <- beta[sel] betap <- sort(betap) betas <- betap[1] sel <- beta < 0 betan <- beta[sel] nn <- length(betan) betan <- sort(betan) betal <- betan[nn] zl <- qnorm((nn - 0.375)/(p + 0.25)) zs <- qnorm((nn + 1 - 0.375)/(p + 0.25)) gap <- ((betas - betal)/psehe)/(zs - zl) return(gap) }