Computes the 3-sigma control limits (lower and upper) for an
FLQuant time series using the average moving range method
(Montgomery, 2009), and performs a non-parametric runs test to assess
whether the series shows a systematic trend.
Usage
sigma3(x, mixing = "less", type = "residual")
Arguments
- x
An FLQuant time series (typically residuals).
- mixing
Character; alternative hypothesis for the runs test. One
of "two.sided", "less" (default, tests for trend), or
"greater".
- type
Character; reserved for future use; currently ignored.
Value
A named list with elements:
- lcl
Lower 3-sigma control limit (negative).
- ucl
Upper 3-sigma control limit (positive).
- p.value
p-value from the runs test; values >= 0.05 suggest
acceptable randomness.
Examples
data(ple4)
sigma3(catch.n(ple4))
#> $lcl
#> [1] -90490.17
#>
#> $ucl
#> [1] 90490.17
#>
#> $p.value
#> [1] 0.001
#>
data(nsher)
sigma3(residuals(nsher))
#> $lcl
#> [1] -1.189247
#>
#> $ucl
#> [1] 1.189247
#>
#> $p.value
#> [1] 0.234
#>
