Skip to contents

Create a time series of decline data

Usage

one_box(x, ini, ..., t_end = 100, res = 0.01)

# S3 method for numeric
one_box(x, ini = 1, ..., t_end = 100, res = 0.01)

# S3 method for character
one_box(x, ini = 1, parms, ..., t_end = 100, res = 0.01)

# S3 method for mkinfit
one_box(x, ini = "model", ..., t_end = 100, res = 0.01)

Arguments

x

When numeric, this is the half-life to be used for an exponential decline. When a character string specifying a parent decline model is given e.g. FOMC, parms must contain the corresponding parameters. If x is an mkinfit object, the decline is calculated from this object.

ini

The initial amount. If x is an mkinfit object, and ini is 'model', the fitted initial concentrations are used. Otherwise, ini must be numeric. If it has length one, it is used for the parent and initial values of metabolites are zero, otherwise, it must give values for all observed variables.

...

Further arguments passed to methods

t_end

End of the time series

res

Resolution of the time series

parms

A named numeric vector containing the model parameters

Value

An object of class one_box, inheriting from ts.

Examples

# Only use a half-life
pred_0 <- one_box(10)
plot(pred_0)


# Use a fitted mkinfit model
require(mkin)
fit <- mkinfit("FOMC", FOCUS_2006_C, quiet = TRUE)
pred_1 <- one_box(fit)
plot(pred_1)


# Use a model with more than one observed variable
m_2 <- mkinmod(parent = mkinsub("SFO", "m1"), m1 = mkinsub("SFO"))
#> Temporary DLL for differentials generated and loaded
fit_2 <- mkinfit(m_2, FOCUS_2006_D, quiet = TRUE)
#> Warning: Observations with value of zero were removed from the data
pred_2 <- one_box(fit_2, ini = "model")
plot(pred_2)