cross_validate: Time series cross-validation with generalized Bayesian INAR...

In bayesianfactory/BayesINAR: BayesINAR model package

Description

Obtain predictions of a given test set and the mean absolute errors

Usage

cross_validate(time_series, model = "inar", p = 1, h = 1,
  training_epoch = NULL, prior = list(a_alpha = NULL, a0 = NULL, b0 =
  NULL, tau = NULL, k0 = NULL, a_tau = NULL, b_tau = NULL, a_w = NULL, b_w
  = NULL, a_theta = NULL, b_theta = NULL, sigma = NULL, lambda_max = NULL),
  burn_in = 10^3, chain_length = 10^4, random_seed = 1761,
  verbose = TRUE)

Arguments

time_series	A univariate time series.
h	Number of steps ahead to be predicted.
training_epoch	The last observation of the first training set.
prior	List of prior hyperparameters where: a_alpha Hyperparameters of the thinning component. a_theta, b_theta Hyperparameters of the Geometric innovation parameter. a_lambda, b_lambda Hyperparameters of the Poisson innovation rate. a_w, b_w Hyperparameters of the Geometric-Poisson mixture weight. a_tau, b_tau Hyperparameters of the concentration parameter Gamma prior. a0, b0 Base measure hyperparameters. lambda_max Hyperparameter of the uniform distribution that minimizes the corresponding D-KL.
burn_in	Number of iterations for the "burn-in" period which are discarded in the chain.
chain_length	Number of iterations of the chain.
random_seed	Value of the random seed generator.
verbose	If TRUE log info is provided.

Value

A list with the following elements:

est

Predictions of the test set.

mae

Mean Absolute Error of the test set predictions.

bayesianfactory/BayesINAR documentation built on Dec. 16, 2019, 12:38 a.m.