GlmResponse: Generate a Generalized Linear Model (GLM) Response model
In mcplR: Multiple Cue Probability Learning in R

Description Usage Arguments Details Value Author(s)

Constructs a GlmResponse object.

1
2
3

GlmResponse(formula,family=gaussian(),data,covariate,
combine=function(x,z) { cbind(x,z) },parameters=list(),ntimes=NULL,
replicate=TRUE,fixed,base=NULL,parStruct,subset)

`formula`	an object of class `formula` (or one that can be coerced to that class): a symbolic description of the model to be fitted. For more details of model specification, see `lm` or `glm`.
`family`	a `family` object. For more details, see `glm`.
`data`	(optional) data frame.
`covariate`	(optional) a matrix or a formula specifying additional covariates. See details.
`combine`	a function taking two arguments which produces a new predictor matrix for the GlmModel. The first argument is the prediction from a LearningModel and the second argument is matrix specified through the covariate argument.
`parameters`	an (optional) named list with (starting) values of the parameters. If no values are supplied, defaults are used.
`ntimes`	an optional vector with, for each repetition in the data, the total number of trials.
`replicate`	logical to indicate whether model parameters are identical for each replication in ntimes.
`fixed`	a logical vector indicating whether model parameters are fixed or free
`base`	assign one of the levels of the criterion variable the role of base category.
`parStruct`	a ParStruct object. If supplied, the fixed argument above will be ignored
`subset`	(optional) subset.

The GlmResponse function sets up a GLM to use as ResponseModel. If the returned GlmResponse is to be used as part of an McplModel, then the formula should contain an entry for the predictions of the learningModel in that McplModel. When fitting the McplModel, the predictions of the learningModel at the current iteration will be used. It is therefore necessary to give the name of the predictions of the learning model in the formula, so that this variable can be replaced with the current variable in the estimation. Ideally, the variable specified by learnmmodpred should not be contained in the data.frame specified by data.

A (fitted) object of class GlmResponse.

Maarten Speekenbrink

mcplR documentation built on May 2, 2019, 4:42 p.m.