plot.dynamiteformula {dynamite}

Plot the Model Structure as a Directed Acyclic Graph (DAG)

Description

Plot a snapshot of the model structure at a specific time point with a window of the highest-order lag dependency both into the past and the future as a directed acyclic graph (DAG). Only response variables are shown in the plot. This function can also produce a TikZ code of the DAG to be used in reports and publications.

Usage

## S3 method for class 'dynamiteformula'
plot(
  x,
  show_auxiliary = TRUE,
  show_covariates = FALSE,
  tikz = FALSE,
  vertex_size = 0.25,
  label_size = 18,
  ...
)

Arguments

x

[dynamiteformula]
The model formula.

show_auxiliary

[logical(1)]
Should deterministic auxiliary responses be shown in the plot? If FALSE, the vertices corresponding to such responses will be projected out. The default is TRUE.

show_covariates

[logical(1)]
Should unmodeled covariates be shown in the plot? The defaults is FALSE.

tikz

[logical(1)]
Should the DAG be returned in TikZ format? The default is FALSE returning a ggplot object instead.

vertex_size

[double(1)]
The size (radius) of the vertex circles used in the ggplot DAG. (The vertical and horizontal distances between vertices in the grid are 1, for reference.)

label_size

[double(1)]
Font size (in points) to use for the vertex labels in the ggplot DAG.

...

Not used..

Value

A ggplot object, or a character string if tikz = TRUE.

See Also

Drawing plots plot.dynamitefit()

Examples

data.table::setDTthreads(1) # For CRAN
multichannel_formula <- obs(g ~ lag(g) + lag(logp), family = "gaussian") +
  obs(p ~ lag(g) + lag(logp) + lag(b), family = "poisson") +
  obs(b ~ lag(b) * lag(logp) + lag(b) * lag(g), family = "bernoulli") +
  aux(numeric(logp) ~ log(p + 1))
# A ggplot
plot(multichannel_formula)
# TikZ format
plot(multichannel_formula, tikz = TRUE)