| pen.log.like {pendensity} | R Documentation |
Calculating the log likelihood
Description
Calculating the considered log likelihood. If one chooses lambda0=0, one gets the (actual) unpenalized log likelihood. Otherwise, one gets the penalized log likelihood for the used fitted values of the response y and the actual parameter set beta.
Usage
pen.log.like(penden.env,lambda0,f.hat.val=NULL,beta.val=NULL)
Arguments
penden.env |
Containing all information, environment of pendensity() |
lambda0 |
penalty parameter lambda |
f.hat.val |
matrix contains the fitted values of the response, if NULL the matrix is caught in the environment |
beta.val |
actual parameter set beta, if NULL the vector is caught in the environment |
Details
The calculation depends on the fitted values of the response as well as on the penalty parameter lambda and the penalty matrix Dm.
l(\beta)=\sum_{i=1}^{n} \left[ \log \{\sum_{k=-K}^K c_k(x_i,\beta) \boldsymbol\phi_k(y_i)\} \right]- \frac 12 \lambda \beta^T D_m \beta
.
The needed values are saved in the environment.
Value
Returns the log likelihood depending on the penalty parameter lambda.
Author(s)
Christian Schellhase <cschellhase@wiwi.uni-bielefeld.de>
References
Density Estimation with a Penalized Mixture Approach, Schellhase C. and Kauermann G. (2012), Computational Statistics 27 (4), p. 757-777.