dLR_nb: A function that returns the value of the partial derivative...

In CenterForStatistics-UGent/RCM: Fit row-column association models with the negative binomial distribution for the microbiome

A function that returns the value of the partial derivative of the log-likelihood ratio to alpha, keeping the response functions fixed

Description

A function that returns the value of the partial derivative of the log-likelihood ratio to alpha, keeping the response functions fixed

Usage

dLR_nb(
  Alpha,
  X,
  CC,
  responseFun = c("linear", "quadratic", "nonparametric", "dynamic"),
  psi,
  NB_params,
  NB_params_noLab,
  d,
  alphaK,
  k,
  centMat,
  nLambda,
  nLambda1s,
  thetaMat,
  muMarg,
  ncols,
  envGradEst,
  allowMissingness,
  naId,
  ...
)

Arguments

Alpha	a vector of length d + k*(2+(k-1)/2), the environmental gradient plus the lagrangian multipliers
X	the n-by-p count matrix
CC	a n-by-d covariate vector
responseFun	a character string indicating the type of response function
psi	a scalar, an importance parameter
NB_params	Starting values for the NB_params
NB_params_noLab	Starting values for the NB_params without label
d	an integer, the number of covariate parameters
alphaK	a matrix of environmental gradients of lower dimensions
k	an integer, the current dimension
centMat	a nLambda1s-by-d centering matrix
nLambda	an integer, number of lagrangian multipliers
nLambda1s	an integer, number of centering restrictions
thetaMat	a matrix of size n-by-p with estimated dispersion parameters
muMarg	an n-by-p offset matrix
ncols	a scalar, the number of columns of X
envGradEst	a character string, indicating how the environmental gradient should be fitted. 'LR' using the likelihood-ratio criterion, or 'ML' a full maximum likelihood solution
allowMissingness	A boolean, are missing values present
naId	The numeric index of the missing values in X
...	further arguments passed on to other methods

Value

: The value of the lagrangian and the constraining equations

CenterForStatistics-UGent/RCM documentation built on April 24, 2023, 8:26 p.m.