generateWeather: Generate multivariate weather series using a Bayesian Network...

In MNLR/BNWeatherGen: Weather Generators with Bayesian Networks

Generate multivariate weather series using a Bayesian Network Weather Generator.

Description

Uses a Bayesian network model, output from buildWeatherGenerator() to simulate a multivariate series. If the model was learnt without predictors, x must be NULL. Otherwise x is required.

Usage

generateWeather(
  wg,
  initial = NULL,
  n = 1,
  x = NULL,
  inference.type = NULL,
  initial.date = NULL,
  advance.type = "simulation",
  threshold.vector = 0.5,
  resample.size = 10000,
  event = "1"
)

Arguments

wg	Bayesian network weather generator model, as output from buildWeatherGenerator()
initial	An initial observation, as vector. If NULL, a random observation will be chosen.
n	Number of observations to generate.
x	A predictors object (grid)
inference.type	Either "approximate" or "exact". If junction tree is unavailable, exact inference is deactivated
initial.date	A date element. If provided, rownames will be labeled after this date.
advance.type	"simulation" (default) is used to generate stochastic weather series. Other options are included for debugging purposes, which are "exact", so that the series are generated deterministically using most likely and threshold.vector.
threshold.vector	Ignored if advance.type = "simulation", used as the threshold for deciding if probability value should be 1 or 0.
event	Name of the positive event.

MNLR/BNWeatherGen documentation built on June 2, 2023, 9:02 p.m.