nlreg.object: Nonlinear Heteroscedastic Model Object
In hoa: Higher Order Likelihood Inference

Description Arguments Generation Methods Note See Also

Class of objects returned when fitting a nonlinear heteroscedastic model.

The following components must be included in a nlreg object:

`coef`	the MLEs of the regression coefficients, that is, of the parameters appearing in the right-hand side of the `formula` argument in the call that generated the `nlreg` object.
`varPar`	the MLEs of the variance parameters appearing in the `weights` argument of the call that generated the `nlreg` object. If this argument was missing, the MLE of \code{log(s^2)}, the logarithm of the constant variance, is returned.
`offset`	a numerical vector with a single named element indicating the parameter of interest and the value to which it was fixed while fitting the nonlinear model.
`logLik`	the log likelihood from the fit.
`meanFun`	the formula expression of the mean function.
`varFun`	the formula expression of the variance function.
`data`	a list representing a summary of the original data with the following components: `'offset name'` the predictor variable with no duplicated value. `repl` the number of replicates available for each value of the predictor. `dupl` a vector of the same length than the predictor variable indicating the position of each data point in the offset name component. `t1` the sum of the reponses for each design point in the offset name component. `t2` the sum of the squared responses for each design point in the offset name component.
`fitted`	the fitted values, that is, the mean function evaluated at each data point.
`weights`	the variance function evaluated at each data point.
`residuals`	the response/standardized residuals from the fit.
`start`	the starting values used to initialize the fitting routine.
`call`	an image of the call to `nlreg`, but with all the arguments explicitly named.
`ws`	a list containing information that is used in subsequent calculations with the following components: `allPar` the MLEs of all parameters. `homVar` a logical value indicating whether the variance function is constant. `xVar` a logical value indicating whether the variance function depends on the predictor variable. `hoa` the value of the `hoa` argument in the call that generated the `nlreg` object. `missingData` a logical value indicating whether the `data` argument was missing in the call that generated the `nlreg` object. `frame` the name of the data frame if specified in the call to `nlreg`. `iter` the number of iteration required until convergence (only for non constant variance function). `md` a function definition that returns the first two derivatives of the mean function if `hoa = TRUE` in the function call that generated the `nlreg` object. `vd` a function definition that returns the first two derivatives of the variance function if `hoa = TRUE` in the function call that generated the `nlreg` object.

This class of objects is returned by the nlreg function to represent a fitted nonlinear heteroscedastic model. Class nlreg inherits from class nls, which represents a homoscedastic nonlinear model fit.

Objects of this class have methods for the functions print, summary, fitted among others.

The residuals, fitted values and coefficients should be extracted by the generic functions of the same name, rather than by the $ operator.

The data and ws components are not intended to be directly accessed by users, but rather contain information invoked by functions such as profile and nlreg.diag.

nlreg, nls

hoa documentation built on May 2, 2019, 8:56 a.m.