Executes a hill-climbing algorithm to iteratively improve a candidate data set by minimizing a supplied error function. Supports both the LM and LME modules.
Usage
hill_climb(
current_candidate,
error_function,
N,
hill_climbs = 100,
LME = FALSE,
num_preds = NULL,
progress_bar = TRUE,
neighborhood_size = 4,
w.length = NULL,
outcome = NULL,
progressor = NULL,
pb_interval = NULL
)Arguments
- current_candidate
A data frame representing the initial candidate solution to be optimized.
- error_function
An objective function that takes a candidate and returns a list containing element `$total_error`.
- N
Integer. Sample size; the number of subjects in `current_candidate`.
- hill_climbs
Integer. Maximum number of iterations for hill climbing. Default is 1e2.
- LME
Logical. If TRUE, perform moves appropriate for mixed-effects data (long-to-wide swaps). Default is FALSE.
- num_preds
Integer. Number of predictors (columns) when `LME = FALSE`. Required if not mixed-effects.
- progress_bar
Logical. Whether to display a text progress bar. Default is TRUE.
- neighborhood_size
Integer. Number of candidate moves evaluated per iteration. Default is 4.
- w.length
Integer. Number of columns in the wide-format data when `LME = TRUE`. Required for mixed-effects.
- outcome
Optional vector. Outcome variable for standard model moves (unused if error_function handles it).
- progressor
Optional function. Callback for external progress updates (internal use).
- pb_interval
Optional numeric. Interval (in iterations) between progressor calls.
Value
A list with components:
- best_candidate
The optimized candidate structure achieving lowest error.
- best_error
Numeric. The minimum value of the objective function found during optimization.
Examples
if (FALSE) { # \dontrun{
hill_climb(
current_candidate = data.frame(),
outcome = NULL,
N = 100,
error_function = function(candidate) {},
hill_climbs = 100,
LME = TRUE,
num_preds = NULL,
progress_bar = TRUE,
progressor = NULL,
pb_interval= NULL)
} # }