GSDTuneJob
GSDTuneJob
An R6 class generator. Use $new() to create a GSDTuneJob object.
targetTarget design function name ("gsDesign" or "gsSurv").
base_argsNamed list of fixed arguments passed to the target function.
tune_specsNamed list of tuning specifications for explored arguments.
param_spaceInternal parameter space used for configuration generation.
specAudit record including base/tuned args and sessionInfo().
new()Create a new tune job.
GSDTuneJob$new(target = c("gsDesign", "gsSurv", "gsSurvCalendar"), args)targetTarget function name ("gsDesign", "gsSurv", or "gsSurvCalendar").
argsNamed list of evaluated arguments.
run()Evaluate configurations under a search strategy.
GSDTuneJob$run(
strategy = c("grid", "random"),
n = NULL,
parallel = TRUE,
seed = NULL,
cache_dir = NULL,
metrics_fun = NULL
)strategySearch strategy ("grid" or "random").
nNumber of configurations for random search.
parallelWhether to evaluate configurations in parallel.
seedOptional seed for reproducibility.
cache_dirOptional directory to cache design objects as RDS.
metrics_funOptional metric hook.
results()Return the results data.frame.
GSDTuneJob$results()
summarize()Summarize the run (counts + numeric metric summaries).
GSDTuneJob$summarize()
design()Retrieve a design object for configuration i.
GSDTuneJob$design(i)iRow index of the configuration.
call_args()Return the underlying argument list for configuration i.
GSDTuneJob$call_args(i)iRow index of the configuration.
best()Rank configurations by a metric (with optional constraints).
GSDTuneJob$best(metric, direction = c("min", "max"), constraints = NULL)metricMetric column name.
directionRanking direction ("min" or "max").
constraintsOptional constraints (function or expression).
pareto()Compute a Pareto (non-dominated) set for multiple metrics.
GSDTuneJob$pareto(metrics, directions)metricsMetric column names.
directionsDirections for each metric ("min"/"max").
plot()Create a quick exploration plot.
GSDTuneJob$plot(metric, x, color = NULL, facet = NULL)metricY-axis metric column name.
xX-axis column name.
colorOptional color column name.
facetOptional faceting column name.
report()Render an HTML report.
GSDTuneJob$report(path)pathOutput HTML file path.
clone()The objects of this class are cloneable with this method.
GSDTuneJob$clone(deep = FALSE)deepWhether to make a deep clone.
R6 class representing a dependency-aware tuning job for group sequential
designs created by gsDesign::gsDesign() or gsDesign::gsSurv().
job <- GSDTuneJob$new(target = "gsDesign", args = list(k = 3, alpha = 0.025))
job$spec$target
Run the code above in your browser using DataLab