Skip to contents

Generate Data-Driven Candidate Values

Source: R/auto_tuning.R
generate_candidate_values.Rd

Generates candidate values for small.cell.reduction parameter using data-driven approaches including percentile analysis, matrix sparsity assessment, and network density optimization.

Usage

generate_candidate_values(
  mx,
  method = "hybrid",
  max_candidates = 20,
  include_zero = TRUE
)

Arguments

mx

Mobility matrix for analysis.

method

Character string specifying generation strategy:

  • "percentile": Percentile-based thresholds from data distribution

  • "sparsity": Matrix sparsity-informed selection

  • "density": Network density optimization targets

  • "hybrid": Combination of multiple strategies (default)

max_candidates

Maximum number of candidates to generate. Default is 20.

include_zero

Logical indicating whether to include 0 as a candidate. Default is TRUE.

Value

Numeric vector of candidate small.cell.reduction values, sorted in ascending order.

Details

Percentile Method: Generates thresholds based on the distribution of non-zero values in the mobility matrix. Uses quantiles from 5

Sparsity Method: Analyzes matrix sparsity patterns and generates candidates that preserve different levels of network connectivity.

Density Method: Targets specific network density levels (0.1, 0.2, etc.) by estimating required thresholds.

Hybrid Method: Combines all approaches and selects diverse, non-redundant candidates covering the full parameter space.

See also

auto_tune_small_cell_reduction

Examples

if (FALSE) { # \dontrun{
# Generate sample data
mobility_data <- generate_mobility_data(n_classes = 5, seed = 123)

# Percentile-based candidates
candidates_pct <- generate_candidate_values(mobility_data, "percentile")

# Sparsity-informed candidates  
candidates_sparse <- generate_candidate_values(mobility_data, "sparsity")

# Hybrid approach (recommended)
candidates_hybrid <- generate_candidate_values(mobility_data, "hybrid")

print(candidates_hybrid)
} # }