maxmult {hyperdirichlet} | R Documentation |
Gives a maximum likelihood estimate for the parameters of a Dirichlet distribution, on the basis of datapoints drawn from a multivariate beta distribution
maxmult(M, start_a = NULL, give = FALSE, method = "nlm", ...)
M |
Integer matrix whose rows give multinomial observations |
start_a |
Start point for optimization, with default NULL
being interpreted as Mosimann's formula 29 |
give |
Boolean, with default FALSE meaning to return just
the point estimate and TRUE meaning to return all the output
from the optimization routine |
method |
Text string specifying the optimization routine to use.
Two values coded: default nlm means to use
nlm() and optim meaning to use optim() ;
anything else means to return Mosimann's estimate (equation 29) |
... |
Further arguments passed to nlm() or optim() |
Finds the maximum likelihood estimate from the equation 7 of Mosimann 1962.
The nlm()
function appears to be better suited to this problem
than optim()
.
Robin K. S. Hankin
J. E. Mosimann 1962. “On the compound multinomial distribution, the multivariate beta-distribution, and correlations among proportions”. Biometrika, volume 49, numbers 1 and 2, pp65-82.
data(pollen) maxmult(pollen, start_a=c(51.81, 0.987, 5.332, 1.961))