Summarize fitted subspace learner

Provides a comprehensive summary of a trained subspace learner, including task information, fitted coefficients, optimization status, and outlier configurations. Prints formatted tables to the console and invisibly returns the summary information.

Usage

# S3 method for class 'LearnerSubspace'
summary(object, ...)

Arguments

object

A trained LearnerSubspace object (or subclass) with fitted subspace parameters

...

Additional arguments (currently unused)

Value

Invisibly returns a list with components:

• status: data.table with optimization status, objective values, number of violations, and observation counts

• coefficients: data.table with fitted subspace parameters (format depends on learner type)

• outliers: data.table with outlier configurations (empty if no outliers or lambda = NULL)

Details

Printed Output:

The function prints three formatted tables:

1. Summary Table:

• Target measure being optimized

• Numeric hyperparameters included in subspace

• Categorical hyperparameters (if any)

2. Coefficients Table:

• Box: min/max bounds per hyperparameter

• Polygon/Ellipsoid: A matrices and b vectors in list columns

• Separate rows for each categorical level (if applicable)

3. Status Table:

• status: Solver convergence status (e.g., "optimal", "solved")

• objective_value: Final objective function value

• n_violations: Number of configurations treated as outliers

• observations: Number of top configurations used for fitting

Status Values:

Fields may be NULL when:

• lambda = NULL: Simple min/max fitting (Box learner only)

• No outliers: All configurations fit within subspace

Dependencies:

Requires knitr package for formatted table output. If not available, falls back to basic printing.

See also

LearnerSubspace for the base learner class. Methods coef() and augment() for extracting fitted parameters. outliers for extracting outlier configurations.

Examples

if (FALSE) { # \dontrun{
# Train learner
task <- TaskSubspace$new(data, target_measure = "auc",
                         hps = c("learning_rate", "max_depth"))
learner <- LearnerSubspaceBox$new(task)
learner$train(q_val = 0.9, lambda = 0.1)

# Print summary to console
summary(learner)

# Capture summary information
info <- summary(learner)
print(info$status)
print(info$coefficients)
print(info$outliers)

# With categorical hyperparameters
task <- TaskSubspace$new(data, target_measure = "auc",
                         hps = c("learning_rate", "max_depth"),
                         cat_hps = "optimizer")
learner <- LearnerSubspaceBox$new(task)
learner$train(q_val = 0.9, lambda = 0.1)
summary(learner)  # Separate status rows per optimizer
} # }