cr.exp.loss: Break down expected future optimality into two components: 1)...

In mst1g15/biasedcoin:

Description

Break down expected future optimality into two components: 1) assuming that the current response is 0 2) assuming that the current response is 1 Find the weighted average of the two cases

Usage

cr.exp.loss(z.now, t.now, zp, N, design, k, beta, y, cr, wr, prior.scale, bayes,
  dyn = NULL, Nprobs = NULL, ...)

Arguments

z.now	vector of covariate values for current unit
t.now	treatment of current unit
zp	vector of probabilities for each level of covariate z (needs to in the same order as all.z below)
N	natural number greater than 0 for horizon
design	design matrix constructed for all units up until the current unit
k	number of covariates
beta	vector of current estimates for regression coefficients
y	vectir of responses generated up until current unit
cr	matrix of contrasts
wr	matrix of weights
prior.scale	prior scale parameter
bayes	set to T if bayesglm is used instead of glm. Default prior assumed
dyn	set to NULL of there are no dynamic covariates, set to T if there are dynamic covariates
Nprobs	a counter to be used if there are dynamic covariates
...	further arguments to be passed to <lossfunc>

Value

expected value of objective function one step ahead in the future

mst1g15/biasedcoin documentation built on Nov. 26, 2019, 4:01 a.m.