fit_model: Internal regression functions

In Laksafoss/ICPSurv: Invariant Causal Prediction for Survival Data

Description

The fit_model function is a generic function meant for internal use in the ICPSurv package, and as such it is not exported to the namespace.

Usage

fit_model(method, Y, X, subset, ...)

## Default S3 method:
fit_model(method, Y, X, subset = NULL, ...)

## S3 method for class 'lm'
fit_model(method, Y, X, subset = NULL, ...)

## S3 method for class 'glm'
fit_model(method, Y, X, subset = NULL, ...)

## S3 method for class 'ph'
fit_model(method, Y, X, subset = NULL, ...)

## S3 method for class 'ah'
fit_model(method, Y, X, subset = NULL, ...)

## S3 method for class 'hazard'
fit_model(method, Y, X, subset = NULL, ...)

fit_nonparam_model(method, Y, X, subset, ...)

## Default S3 method:
fit_nonparam_model(method, Y, X, subset = NULL, ...)

## S3 method for class 'ph'
fit_nonparam_model(method, Y, X, subset = NULL, id = NULL, ...)

## S3 method for class 'ah'
fit_nonparam_model(method, Y, X, subset = NULL, id = NULL, ...)

Arguments

method	a method object created by the method_obj function.
Y	a vector or Surv object describing the target variable.
X	a matrix, vector or data frame describing the covariates.
subset	an optional vector specifying a subset of observations to be used in the fitting process.
...	additional arguments to be passed to lower level functions.
id	for timevarying covariates the variable must associate each record with the id of a subject.

Details

The fit_model.X and fit_nonparam_model.X functions are internal fitter functions. They are usually all quite simple as their main functionality apart from fitting the model is ensureing that the output is compatible with the plausible_predictor_tests.

All fit_model methods must return the following:

• coefficients The estimated regression coefficients.

• deviance The deviance of the fitted model, i.e. -2 log(likelihood)

• df Degrees of freedom in fitted model.

• scale Scale.

All fit_nonparam_model methods must return the following:

• cum Cumulative regression coefficient.

• cum.var Cumulative variance of regression effects.

Morover if n.sim in the method is non-zero then the fit_nonparam_model also returnes the following

• sup Kolmogorov–Smirnov test.

• int Cramér–von Mises test.

Value

Both the fit_model and fit_nonparam_model methods return a list.

fit_model methods must return the following:

coefficients	The estimated regression coefficients.
deviance	The deviance of the fitted model, i.e. -2 log(likelihood)
df	Degrees of freedom in fitted model.
scale	Scale.

fit_nonparam_model methods return the following:

cum	Cumulative regression coefficient.
cum.var	Cumulative variance of regression effects.
sup	Kolmogorov–Smirnov test (only returned if n.sim not zero).
int	Cramér–von Mises test (only returned if n.sim not zero).

See Also

ICP, plausible_predictor_test for the full wrapper functions.

Laksafoss/ICPSurv documentation built on Feb. 26, 2020, 11:32 a.m.