fitLNBzt: ML estimation of (zero-truncated) LNB

In schmettow/LNB: (zero- truncated) logit-normal binomial distribution functions

Description

takes a margin sum data set and the size (number of sessions) and fits by ML for extreme sparse or strongly dispersed data, the fit may fail. Experiment with the start values (startval) in that case.

Usage

fitLNBzt(ms,n,startval=c(-1,2))

Arguments

ms	vector with margin sum
n	number of trials
K	starting values for mu and sd

Details

takes a margin sum data set and the size (number of sessions) and fits by ML for extreme sparse or strongly dispersed data, the fit may fail. Experiment with the start values (startval) in that case.

Value

fitLNB

mu	mean of logitnormal prior
sd	standard deviation of logitnormal prior
n	number of sessions/trials
discovered	number of responses > 0
ms	margin sum
nlogLik	negative log-likelihood
AIC	Akaike Information Criterion
zt	zero-truncated estimation (logical, TRUE with fitLNBzt)

Author(s)

Martin Schmettow

References

Schmettow, M. (2009). Controlling the usability evaluation process under varying defect visibility. In BCS HCI 09: Proceedings of the 23rd British HCI Group Annual Conference on People and Computers: Celebrating People and Technology (pp. 188-197). Swinton, UK: British Computer Society.

See Also

dlnbinom.zt

Examples

  ms <- rlnbinom(100, 15, 0.5, 1.2)
  ms <- ms[ms > 0]
  fit <- fitLNBzt(ms, 15)
  fit
  summary(fit)
  dhat(fit)
  dnull(fit)
  plot(fit)

schmettow/LNB documentation built on May 29, 2019, 3:41 p.m.