Simulations: Function to simulate compositional time series data

In composits: Compositional, Multivariate and Univariate Time Series Outlier Ensemble

Function to simulate compositional time series data

Description

Function to simulate compositional time series data

Usage

Simulations(N, TT, K, A, B, C, mu, D, outliers_discre, q)

Arguments

N	The number of categories in the composition
TT	The time series length
K	The state vector dimension
A	The N x K matrix of factor loadings in the observation equation
B	The K x K autoregressive matrix of the transition equation
C	The K x K matrix determining the magnitude of the persistent outliers
mu	The K-dimensional intercept vector in the transition equation
D	A K x K matrix determining the variance-covariance matrix of the error term
outliers_discre	An R x 3 matrix of discretionary outliers. R denotes the number of discretionary outliers. The first, second and third columns denote the time position, the composite position and the magnitude of the outliers
q	Probability of persistent outlier eventuating

Value

A list with the following components:

datasim	A TT x K data frame with the generated time series compositional data.
outliers_persist	A matrix indicating the time location of the persistant outliers (first column) and the factors (or states) where the outlier eventuates (second column).
outliers_discre	A matrix equivalent to the function argument provided by the user.
outliers_timeloc	A vector with the time location of all the outliers.

Examples

set.seed(2000)
N <- 30
K <- 2
TT <- 500
A <- matrix(rnorm(N*K, 0, 0.3), N, K)
B <- matrix(c(0.8,0,0,0.5), K, K)
C <- matrix(c(5,0,0,4), K, K)
mu <- c(0.3, 0.7)
D <- matrix(c(0.4,0,0,0.4), K, K)
outliers_discre <- matrix(c(117, 2, 10, 40, 8, 200), 2, 3, byrow = TRUE)
q <- 0.005
y <-  Simulations(N = N,
                 TT = TT,
                 K = K,
                 A = A,
                 B = B,
                 C = C,
                 mu = mu,
                 D = D,
                 outliers_discre = outliers_discre,
                 q = q)

composits documentation built on May 25, 2022, 9:11 a.m.