generate.data: Generate data from MPT models.
In MPTinR2: Analyze Multinomial Processing Tree Models

Description Usage Arguments Details Value Author(s) See Also

generate.data in MPTinR is a generic function that generates random data or predicted data for MPT models from given parameter values.

## S4 method for signature 'mpt.model'
generate.data(model, parameter.values, n.per.tree, n = 1, random = TRUE, ...)
## S4 method for signature 'mpt'
generate.data(model, n = 1, random = TRUE, ...)

`model`	Either a MPT model (an object of class `bmpt.model` or `mpt.model`, returned from `make.mpt`) or a fitted MPT model (object of class `bmpt` or `mpt`, returned from `fit.mpt`)
`parameter.values`	named vector or matrix with the parameter estiamtes to be used for data generation. If a vector, `names(parameter.values)` must correspond to the parameters in the model. If a matrix, each row represents one set of parameter estimates and the `colnames(parameter.values)` must correspond to the parameters in the model.
`n.per.tree`	vector or matrix with length/ number of columns corresponding to the number of categories in the model. Each value must correspond to the n per tree for that category. If a matrix, each row represents one set of ns.
`n`	numeric. Number of random samples to be drawn per set of parameter values (ignored if `random = FALSE`).
`random`	logical. If `TRUE` (the default) random samples from a multinomial distribution are drawn. If `FALSE` the values predicted from the given parameter estimates are returned (and argument `n` is ignored).
`...`	currently ignored.

This function generates data from either a mutlinomial processing tree (MPT) model object as produced by make.mpt (which is either of class bmpt.model or mpt.model) or a fitted MPT model object as produced by fit.mpt (which is either of class bmpt or mpt).

To generate data from a MPT models you also need to specify parameter values and the desired n. Both can be vectors to generate data from a single individual or a matrix to generate data from multiple individuals with each row corresponding to a single individual. The parameter values need to ba a named object, either a vector or a matrix. The names for a vector or the colnames of a matrix need to correspond to the free parameters in the model. Additional parameters or named objects not part of the model should be ignored.

If random is TRUE, a 3-dimensional array with dim(n, n.categories, n.parameter.sets):

`first dimension`	the number of random draws for each set of parameters (i.e., = `n`).
`second dimension`	the response categories in the model
`third dimension`	the parameter set (i.e., the number of the row in the `parameter.values` object or the dataset)