View source: R/model_selection.R
penalty_BaraudGiraudHuet_likelihood | R Documentation |
penalty_BaraudGiraudHuet_likelihood
is the penalty defined by :
pen' = ntaxa * log(1 + pen/(ntaxa - K)) with
pen = C * (ntaxa - K)/(ntaxa - K - 1) * EDkhi[K + 1; ntaxa - K - 1; exp(-Delta_K)/(K + 1)]
and Delta = log(model_complexity) + log(K + 1)
such that sum(exp(-Delta_K)) < infty.
penalty_BaraudGiraudHuet_likelihood(K, model_complexity, ntaxa, C = 1.1)
K |
the dimension of the model. |
model_complexity |
the complexity of the set of models with dimension K. |
ntaxa |
the number of tips. |
C |
a constant, C > 1. Default is C = 1.1 (as suggested in Baraud Giraud Huet (2009)) |
See Baraud Giraud Huet (2009, 2011).
Must be applied to log-likelihood criterion.
Function pen is computed using function penalty
from package
LINselect
.
value of the penalty.
penalty_BirgeMassart_shape1
,
penalty_BirgeMassart_shape2
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.