set_control {acepack} | R Documentation |
Set internal parameters that control ACE and AVAS algorithms
Description
These parameters are used in the smoothing routines of ACE and AVAS. ACE and AVAS both have their own smoothing implementations. This sets them globally for the package.
The default values are good for the vast majority of cases. This routine is included to provide complete control to the user, but is rarely needed.
Usage
set_control(
alpha = NULL,
big = NULL,
span = NULL,
sml = NULL,
eps = NULL,
spans = NULL,
maxit = NULL,
nterm = NULL
)
Arguments
alpha |
numeric(1); AVAS; Controls high frequency (small span) penalty used with automatic span selection (base tone control). An alpha < 0.0 or alpha > 10.0 results in no effect. Default is 5.0. |
big |
numeric(1); ACE and AVAS; a large floating point number. Default is 1.0e30. |
span |
numeric(1); ACE and AVAS; Span to use in smoothing represents the fraction of observations in smoothing window. Automatic span selection is performed if set to 0.0. Default is 0.0 (automatic). For small samples (n < 40) or if there are substantial serial correlations between observations close in x - value, then a specified fixed span smoother (span > 0) should be used. Reasonable span values are 0.3 to 0.5. |
sml |
numeric(1); AVAS; A small number. Should be set so that '(sml)**(10.0)' does not cause floating point underflow. Default is 1e-30. |
eps |
numeric(1); AVAS; Used to numerically stabilize slope calculations for running linear fits. |
spans |
numeric(3); AVAS; span values for the three running linear smoothers.
Warning: These span values should be changed only with great care. |
maxit |
integer(1); ACE and AVAS; Maximum number of iterations. Default is 20. |
nterm |
integer(1); ACE and AVAS; Number of consecutive iterations for which rsq must change less than delcor for convergence. Default is 3. |
Examples
set_control(maxit=40)
set_control(maxit=20)
set_control(alpha=5.0)
set_control(big=1e30, sml=1e-30)
set_control(eps=1e-3)
set_control(span=0.0, spans=c(0.05, 0.2, 0.5))
set_control(maxit=20, nterm=3)