define.model: Define Stan model

In tkmckenzie/ikde: Iterative Kernel Density Estimation

Description

Defines Stan model and stores input data

Usage

define.model(data, parameters, model, transformed.data = list(),
  transformed.parameters = list())

Arguments

data	A list of data passed to the Stan program. Should be of the form list(data.name = list(type = string, dim = number/string (e.g., "[N]"), value = data.object)).
parameters	A list of parameters used in the Stan program. Should be of the form list(parameter.name = list(type = string, dim = number/string)).
model	A list describing the Stan model. Should be a list with components "priors" and "likelihood".
transformed.data	A list describing data transformations for the Stan program to perform. Should be of the form list(variable.name = list(type = string, dim = number/string, expression = string)).
transformed.parameters	A list describing parameter transformations for the Stan program to perform. Should be of the form list(variable.name = list(type = string, dim = number/string, expression = string)).

Details

Defines inputs to be used for building and eventually fitting Stan model.

Value

Returns an ikde.model object with the following elements

data	A list of data passed to the Stan program
transformed.data	A list describing data transformations for the Stan program to perform
parameters	A list of parameters used in the Stan program
transformed.parameters	A list describing parameter transformations for the Stan program to perform
model	A list describing the Stan model
stan.code	Stan code for the model
stan.data	Data passed to Stan for estimation
stan.dso	DSO for Stan model, allows Stan to run model without recompilation
built	Boolean indicating whether the model has been built
density.variable	List containing two elements: "name" of the variable on which density estimation should be performed on, and "value" indicating the point at which density should be estimated

Examples

data(lm.generated)

X <- lm.generated$X
y <- lm.generated$y

data <- list(N = list(type = "int<lower=1>", dim = 1, value = nrow(X)),
             k = list(type = "int<lower=1>", dim = 1, value = ncol(X)),
             X = list(type = "matrix", dim = "[N, k]", value = X),
             y = list(type = "vector", dim = "[N]", value = y))
parameters <- list(beta = list(type = "vector", dim = "[k]"),
                   sigma_sq = list(type = "real<lower=0>", dim = 1))
model <- list(priors = c("beta ~ normal(0, 10);",
                         "sigma_sq ~ inv_gamma(1, 1);"),
              likelihood = c("y ~ normal(X * beta, sqrt(sigma_sq));"))

ikde.model <- define.model(data, parameters, model)

tkmckenzie/ikde documentation built on May 13, 2019, 9:53 p.m.