Transfer Learning Workflow — tl_transfer

Use unsupervised pre-training before supervised learning: the predictors the formula names are projected onto their principal components, and the supervised model is fitted on the component scores.

Usage

tl_transfer_learning(
  data,
  formula,
  pretrain_method = "pca",
  supervised_method = "tree",
  ...
)

Arguments

data: Training data
formula: Model formula, or a string that parses as one. The PCA is fitted on the numeric predictors it names.
pretrain_method: Pre-training method. Only "pca" is available: predict() has to project new rows, and PCA is the reduction that can.
supervised_method: Supervised learning method (default: "tree", which handles both regression and classification with any number of classes). "logistic" is binary-only and errors on a response with more than two levels.
...: Additional arguments passed to tl_reduce_dimensions, such as n_components

Value

A list with class "tidylearn_transfer" containing:

pretrain_model: The fitted dimensionality reduction model.
supervised_model: The fitted supervised tidylearn model.
formula: The model formula.
method: The supervised learning method used.

Examples

# \donttest{
model <- tl_transfer_learning(iris, Species ~ .,
  pretrain_method = "pca", supervised_method = "tree")
#> Transfer Learning Workflow
#> ==========================
#> [Phase 1] Unsupervised pre-training with pca...
#> [Phase 2] Supervised learning with tree...
# }