Create bootstrapped ensembles of a model
mod_ensemble(model, nreps = 2, data = NULL)
model | a model whose data will be used for resampling |
---|---|
nreps | how many resampling trials should be created |
data | a data table to use for resampling. This is not needed for many common
model types, such as |
The approach to bootstrapping implemented by this function is to create a set of bootstrap
trials all in one go. Then, other functions such as mod_effect()
and mod_eval()
will be
used to extract the information from each of the bootstrap replicates. Many model types in R carry the
data used to train the model as part of the model object produced. For these types of models, e.g. lm
and
glm
, there is no need to provide a value for the data
argument. But there are some
types of models for which the training data cannot be extracted from the model object. In such situations,
you use data =
to provide the data set to use for resampling.