tef_change_power3: Construct a 3-parameter power function of change

In akcochrane/TEfits: Time-evolving model fits

Construct a 3-parameter power function of change

Description

By defining the model variable associated with time (e.g., trial number), and formulas defining each of the nonlinear parameters of time-related change, this function constructs a model that can then be passed to functions for fitting the model (e.g., TEbrm).

Usage

tef_change_power3(
  timeVar,
  parForm = ~1,
  startForm = ~1,
  rateForm = ~1,
  asymForm = ~1,
  rateBase = 2,
  propRemain = 0.25
)

Arguments

timeVar	String. The name of the variable in the model that corresponds to time. The variable of time should be positive and numeric, and the function of change should be expected to happen with increasing time.
parForm	The right-hand side of the formula defining all nonlinear parameters as well as the null [non-time-varying] model.
startForm	The right-hand side of the formula defining the start parameter. Overwrites parForm for this parameter.
rateForm	The right-hand side of the formula defining the rate parameter. Overwrites parForm for this parameter.
asymForm	The right-hand side of the formula defining the asymptote parameter. Overwrites parForm for this parameter.
rateBase	Number. The base of the log (e.g., 2 or exp(1)) of the rate [time constant].
propRemain	Change rate is parameterized in terms of the rateBase log of time to this proportion of change remaining (i.e., 1-propRemain of total change occurs in rateBase^rateParameter time.

Details

Function is under development and is likely to be buggy, and to change frequently.

See Also

TEbrm for examples of how to use this function in specifying models.

Examples

equation_to_fit <- tef_change_power3('timeVar',rateForm = ~ xvar1*xvar2) # both variables should be numeric for TEfit methods! TEbrm should work with factorsas well

equation_to_fit <- tef_change_power3('timeVar'
, parForm = ~ xvar1   # overall parameter formula; overwritten for time-evolving formulas below, leaving this as just the null model
, startForm = ~ xvar2 # start parameter's regression model
, rateForm = ~ (1|participantID)  # rate parameter's [mixed-effects] regression model
, asymForm = ~ xvar3 # asymptote parameter's regression model
)

akcochrane/TEfits documentation built on June 12, 2025, 11:10 a.m.