Rfssa (version 3.1.0)

fssa: Functional Singular Spectrum Analysis (FSSA)

Description

This function performs the decomposition (embedding and functional SVD steps) for univariate (ufssa) or multivariate (mfssa) functional singular spectrum analysis based on the input data type. The input can be a univariate or multivariate functional time series (funts) object.

Usage

fssa(Y, L = Y$N/2, ntriples = 20, type = "ufssa")

Value

An object of class fssa, containing functional objects, eigenvalues, window length, and original data.

Arguments

Y

an object of class funts.

L

a positive integer, the window length, the default is half of FTS length.

ntriples

a positive integer, the number of eigentriples for the decomposition.

type

a string indicating the type of FSSA: "ufssa" (default for univariate FTS) or "mfssa" (default for multivariate FTS).

Examples

data("Callcenter")

# FSSA Decomposition step:
L <- 28
U <- fssa(Callcenter, L)
plot(U, type = "values", d = 10)
plot(U, type = "vectors", d = 4)
plot(U, type = "paired", d = 6)
plot(U, type = "lcurves", d = 4, vars = 1)
plot(U, type = "lheats", d = 4)
plot(U, type = "wcor", d = 10)
plotly_funts(U$Lsingf[[1]])
plot(U$Lsingf[[2]])

if (FALSE) {
#--------------- Multivariate FSSA Example on bivariate -----------------------------
## temperature curves and smoothed images of vegetation
data("Montana")

# MFSSA Decomposition step:
L <- 45
U <- fssa(Montana, L)
plot(U, type = "values", d = 10)
plot(U, type = "vectors", d = 4)
plot(U, type = "lheats", d = 4)
plot(U, type = "lcurves", d = 4, vars = 1)
plot(U, type = "paired", d = 6)
plot(U, type = "periodogram", d = 4)
plot(U, type = "wcor", d = 10)
plotly_funts(U$Lsingf[[1]])
plot(U$Lsingf[[2]])

}