ggplot_descdist_bn: ggplot the posterior distribution of a Bayesian network node...

In Issoufou-Liman/decisionSupportExtra: Miscellaneous Functions for Quantitative Support of Decision Making under Uncertainty

Description

Generate compact plot of Bayesian network node states following the descdist function as customizable ggplot.

Usage

ggplot_descdist_bn(
  bn,
  node,
  boot = 1000,
  obs.col = "darkblue",
  boot.col = "orange",
  title = "Cullen and Frey graph",
  subtitle = node,
  xlab = "square of skewness",
  ylab = "kurtosis",
  obs_geom_size = 4,
  boot_geom_size = 0.02,
  dist_geom_pts_size = 5,
  dist_geom_line_size = 0.6,
  axis_text_size = 12,
  axis_title_size = 12,
  plot_title_size = 20,
  plot_subtitle_size = 17,
  strip_text_size = 18,
  legend_text_size = 12,
  evidence = NULL,
  n_generation = NULL,
  include_relatives = TRUE,
  n_run = 1000
)

Arguments

bn	an object of class bn.fit.
node	character string, the label of the node which conditional distribution is of interest.
boot	If not NULL, boot values of skewness and kurtosis are plotted from bootstrap samples of data. boot must be fixed in this case to an integer above 10.
obs.col	Color used for the observed point on the skewness-kurtosis graph.
boot.col	Color used for bootstrap sample of points on the skewness-kurtosis graph.
title	Title and Subtitle
subtitle	Title and Subtitle
xlab	These are respectively x and y labels.
ylab	These are respectively x and y labels.
obs_geom_size	The size of the geom_point to be used for the empirical distributoion (default to 4), bootstrapping (default to 0.02), theoritical distribution (default to 5), respectively.
boot_geom_size	The size of the geom_point to be used for the empirical distributoion (default to 4), bootstrapping (default to 0.02), theoritical distribution (default to 5), respectively.
dist_geom_pts_size	The size of the geom_point to be used for the empirical distributoion (default to 4), bootstrapping (default to 0.02), theoritical distribution (default to 5), respectively.
dist_geom_line_size	The size of the geom_line to be used for the empirical distributoion. The default is 0.6.
axis_text_size	= 12, Text size respectively corresponding to axis text (default to 12), axis title (default to 12), plot title (default to 20), subtitle (default to 17), strip text (default to 18), and legend (default to 12).
axis_title_size	= 12, Text size respectively corresponding to axis text (default to 12), axis title (default to 12), plot title (default to 20), subtitle (default to 17), strip text (default to 18), and legend (default to 12).
plot_title_size	= 12, Text size respectively corresponding to axis text (default to 12), axis title (default to 12), plot title (default to 20), subtitle (default to 17), strip text (default to 18), and legend (default to 12).
plot_subtitle_size	= 12, Text size respectively corresponding to axis text (default to 12), axis title (default to 12), plot title (default to 20), subtitle (default to 17), strip text (default to 18), and legend (default to 12).
strip_text_size	= 12, Text size respectively corresponding to axis text (default to 12), axis title (default to 12), plot title (default to 20), subtitle (default to 17), strip text (default to 18), and legend (default to 12).
legend_text_size	= 12, Text size respectively corresponding to axis text (default to 12), axis title (default to 12), plot title (default to 20), subtitle (default to 17), strip text (default to 18), and legend (default to 12).
evidence	a name value pair:a named character vector which values are node states and names are node names.
n_generation	how far to go in the network topology for building the conditionning specification for the query?
include_relatives	logical Should parents or ancestors, depending on the the argument n_generation, should be included in the query? If TRUE, the default, these will be internally involved in constructing the evidence argument.
n_run	integer specifying the number of of model run. Default is 1000.

Details

see descdist.

Author(s)

Issoufou Liman

References

Marie Laure Delignette-Muller, Christophe Dutang (2015). fitdistrplus: An R Package for Fitting Distributions. Journal of Statistical Software, 64(4), 1-34. http://www.jstatsoft.org/v64/i04/.

See Also

descdist.

Examples

library (gRain)
library(bnlearn)
## setting a bayesian network with gRain
Soil_type <- cptable (~Soil_type, values = c(0.05, 0.55, 0.4),
levels = c('Sandy', 'Loamy', 'Clayey'))
Manure_application <- cptable(~Manure_application, values = c(0.3, 0.7),
levels = c('FALSE', 'TRUE'))
Soil_water_holding_capacity_tmp <- make_gRain_CPT(
 parent_effects = list(c(0, 2.5, 3), c(0, 2)),
 parent_weights = c(2,1),
 b = 3,
 child_prior = c(0.2,0.5,0.3),
 child_states = c('Low', 'Medium', 'High'),
 parent_states = list(c('Sandy', 'Loamy', 'Clayey'), c('FALSE', 'TRUE'))
)
Soil_water_holding_capacity_values <- Soil_water_holding_capacity_tmp$values
Soil_water_holding_capacity_levels <- Soil_water_holding_capacity_tmp$levels
Soil_water_holding_capacity <- cptable (
~Soil_water_holding_capacity|Soil_type:Manure_application,
values = Soil_water_holding_capacity_values,
levels = Soil_water_holding_capacity_levels)
## Compile conditional probability tables
network <- compileCPT(list(Soil_type, Manure_application, Soil_water_holding_capacity))
## Graphical Independence Network ####
network <- grain(network)
## Use grain object (gRain package)
ggplot_descdist_bn (bn = network, node = 'Soil_water_holding_capacity')
## converting the grain bayesian network to bn.fit
network_bn_fit <- as.bn.fit(network)
## Use bn.fit object (bnlearn package)
ggplot_descdist_bn (bn = network, node = 'Soil_water_holding_capacity')

Issoufou-Liman/decisionSupportExtra documentation built on Dec. 21, 2020, 6:28 p.m.