mfilter: Linear Filtering on a Multivariate Time Series

In timsac: Time Series Analysis and Control Package

Linear Filtering on a Multivariate Time Series

Description

Applies linear filtering to a multivariate time series.

Usage

mfilter(x, filter, method = c("convolution","recursive"), init)

Arguments

x	a multivariate (m-dimensional, n length) time series x[n,m].
filter	an array of filter coefficients. filter[i,j,k] shows the value of i-th row, j-th column, k-th order
method	either "convolution" or "recursive" (and can be abbreviated). If "convolution" a moving average is used: if "recursive" an autoregression is used. For convolution filters, the filter coefficients are for past value only.
init	specifies the initial values of the time series just prior to the start value, in reverse time order. The default is a set of zeros.

Details

This is a multivariate version of "filter" function. Missing values are allowed in 'x' but not in 'filter' (where they would lead to missing values everywhere in the output). Note that there is an implied coefficient 1 at lag 0 in the recursive filter, which gives

y[i,]' =x[,i]' + f[,,1] \times y[i-1,]' + ... +f[,,p] \times y[i-p,]',

No check is made to see if recursive filter is invertible: the output may diverge if it is not. The convolution filter is

y[i,]' = f[,,1] \times x[i,]' + ... + f[,,p] \times x[i-p+1,]'.

Value

mfilter returns a time series object.

Note

'convolve(, type="filter")' uses the FFT for computations and so may be faster for long filters on univariate time series (and so the time alignment is unclear), nor does it handle missing values. 'filter' is faster for a filter of length 100 on a series 1000, for examples.

See Also

convolve, arima.sim

Examples

#AR model simulation
ar <- array(0, dim = c(3,3,2))
ar[, , 1] <- matrix(c(0.4,  0,   0.3,
                      0.2, -0.1, -0.5,
                      0.3,  0.1, 0), nrow = 3, ncol = 3, byrow = TRUE)
ar[, , 2] <- matrix(c(0,  -0.3,  0.5,
                      0.7, -0.4,  1,
                      0,   -0.5,  0.3), nrow = 3, ncol = 3, byrow = TRUE)
x <- matrix(rnorm(100*3), nrow = 100, ncol = 3)
y <- mfilter(x, ar, "recursive")

#Back to white noise
ma <- array(0, dim = c(3,3,3))
ma[, , 1] <- diag(3)
ma[, , 2] <- -ar[, , 1]
ma[, , 3] <- -ar[, , 2]
z <- mfilter(y, ma, "convolution")
mulcor(z)

#AR-MA model simulation
x <- matrix(rnorm(1000*2), nrow = 1000, ncol = 2)
ma <- array(0, dim = c(2,2,2))
ma[, , 1] <- matrix(c( -1.0,  0.0,
                        0.0, -1.0), nrow = 2, ncol = 2, byrow = TRUE)
ma[, , 2] <- matrix(c( -0.2,  0.0,
                       -0.1, -0.3), nrow = 2, ncol = 2, byrow = TRUE)
y <- mfilter(x, ma, "convolution")

ar <- array(0, dim = c(2,2,3))
ar[, , 1] <- matrix(c( -1.0,  0.0,
                        0.0, -1.0), nrow = 2, ncol = 2, byrow = TRUE)
ar[, , 2] <- matrix(c( -0.5, -0.2,
                       -0.2, -0.5), nrow = 2, ncol = 2, byrow = TRUE)
ar[, , 3] <- matrix(c( -0.3, -0.05,
                       -0.1, -0.30), nrow = 2, ncol = 2, byrow = TRUE)
z <- mfilter(y, ar, "recursive")

timsac documentation built on Sept. 30, 2023, 5:06 p.m.