method_obj: Creating a Method Object for ICP
In Laksafoss/ICPSurv: Invariant Causal Prediction for Survival Data

Description Usage Arguments Value See Also Examples

View source: R/ICP-methodobj.R

Creates a method object within the ICP function.

method_obj(model = "glm", method = "EnvirRel", ...)

## S3 method for class 'method_obj'
print(x, exclude = c("id", "tol", "call", "dist"), ...)

`model`	the regression model used in `ICP`. The model class specified here must be interpretable by the generic function `fit_model` or `fit_nonparam_model`. If `model` is not specified it will be set equal to `glm` and `family = "gaussian"` will be added to the method object.
`method`	the testing metod used in `ICP`. The method specified here must be interpretable by the generic function `plausible_predictor_test`. If nothing is specified `method` is set to `EnvirRel`.
`...`	further arguments to be passed to the `plausible_predictor_test`, `fit_model` or `fit_nonparam_model` functions. If `method` is set to "`CR`" then a `solver` can be specified. The standard solver is `QC`. However, if `method` is "`TimeVar`" then `n.sim` can be set to any integer larger then 50. If `model` is `glm` then a `family` should be specified (see `family`). If a `family` is not specified it is set to "`gaussian`".
`x`	a `method_obj` for printing.
`exclude`	variables from the `method_obj` `x` which is not to be included in the printout.

The method_obj function returns a list with all the input parameters and their values. This list has class "method_obj", method and model. If method is CR then the output list also has class equal to solver. If, however, the method is TimeVar then the output also has class nonparamtest.

ICP, plausible_predictor_test, fit_model or fit_nonparam_model for use of the method object.

# The standard method object is a gaussian glm with method EnvirIrrel
ICPSurv:::method_obj()

# A method object for Intersecting Confidence Region analysis of linear models
ICPSurv:::method_obj(model = "lm", method = "CR", solver = "QC", fullAnalysis = TRUE)
# Here 'fullAnalysis' has been set to TRUE which will ensure that QCLP method
# does not use rectangle approximations as a first step