relFit {similarityRichards} | R Documentation |
Calculates the backfitted values (from an 'independent' Richards curve) relative the independent variable (for one sample in one 'experiment').
For each value of the independent variable the (reciprocal) ratio is computed of the (actual) independent value to the independent value - for which the value of the fitted curve (of an other sample) is equal to the mean of the dependent values at the (actual) independent value.
relFit(X, fit = NULL, ny = 0, respName = "SIGNAL", indepName = "ARGX", FUN = NULL, applyFUN2X = function(X, FUN, ...) FUN(X, ...), Smin = -1, Smax = 10, ...)
X |
the dataset with indep and respName
(and for FUN ). |
fit |
a (optional) fit.
summary(fit)$parameters[, "Estimate"]
should return the parameters of a Richards curve. |
ny |
a (optional) numerical value. This value has to given,
if ny is not fitted in the model. |
respName |
a text string with the name of the dependent variable. |
indepName |
a text string with the name of the independent variable. |
FUN |
a function estimating a model for the dataset.
This function is only called if the argument
fit is not given. |
applyFUN2X |
a function applied to the dataset,
retrieving starting value and fixing outliers,
and then calling the function FUN to
estimate the model for the modified dataset.
This function is only called if the argument
fit is not given. |
Smin |
a numeric: lower bound of "linear part". |
Smax |
a numeric: upper bound of "linear part". |
... |
additional arguments. |
The backfitted values relative the actual concentrations are computed as follow (for the one sample in one experiment):
For each value of the independent variable (indepName
)
the observed mean y
of the dependent variable (respName
)
is found from the dataset X
.
For each of these values y
of the dependent variable
the backfitted value is then computed as follows from
the Richards curve with parameters of fit
:
The independent values are found such that
the estimated Richards functions have values y
,
(by the inverse of the Richards functions).
The ratio of the backfitted values to the independent variable
(indepName
) is then finally computed.
"ok.lb" and "ok.ub" are set to 1 if the
independent variable (indepName
) is respectively
above the lower limit or above the upper limit
of the linear part of a 4-parameter logistic curve.
Matrix with relevant values.
Jens Henrik Badsberg
J. L. Sebaugh, P. D. McCray: Defining the linear portion of a sigmoid-shaped curve: bend points. Pharmaceutical Statistics 2003, Volume 2 Issue 3, Pages 167 - 174.