Find change points efficiently in ARIMA(\(p\), \(d\), \(q\)) models

detect_arima() and fastcpd.arima() are wrapper functions of detect() to find change points in ARIMA(\(p\), \(d\), \(q\)) models. Differencing is performed independently inside each candidate segment, so no artificial difference is formed across a proposed change-point boundary. Change-point indices refer to the original, undifferenced series. When \(d = 0\), this is the same model as detect_arma(). The function is similar to detect() except that the data is by default a one-column matrix or univariate vector and thus a formula is not required here.

Usage

detect_arima(data, order = c(1, 1, 0), ...)

fastcpd_arima(data, order = c(1, 1, 0), ...)

fastcpd.arima(data, order = c(1, 1, 0), ...)

Arguments

data

A numeric vector, a matrix, a data frame or a time series object.

order

A vector of length three specifying the order of the ARIMA model.

...

Other arguments passed to detect(), for example, segment_count. The ARIMA-specific include.mean option is accepted here, defaults to FALSE, and must remain FALSE.

Value

A fastcpd object.

See also

detect()

Examples

# \donttest{
small_increments <- rep(c(0.1, -0.1), 20)
large_increments <- rep(c(2, -2), length.out = 41)
x <- c(0, cumsum(c(small_increments, large_increments)))
result <- detect_arima(
  x,
  order = c(0, 1, 0)
)
summary(result)
#> 
#> Call:
#> detect_arima(data = x, order = c(0, 1, 0))
#> 
#> Change points:
#> 41 
#> 
#> Cost values:
#> -33.48908 86.34021 
#> 
#> Parameters:
#>   segment 1 segment 2
#> 1      0.01         4
plot(result)

# }