Skip to contents

Calculate Sweet Spots for Multiple Metrics

Source: R/analysis_compute_sweet_spot.R
sweet_spot.Rd

This function calculates the sweet spots for various metrics over multiple case studies and methods. It determines where the metric differences between two methods cross a reference value.

Usage

sweet_spot(results_freq_df, metrics, based_on_CI = TRUE, nominal_tie = NULL)

Arguments

results_freq_df

A dataframe containing the results frequency data with columns "case_study", "method", "source_denominator_change_factor", "control_drift", and "target_sample_size_per_arm".

metrics

A list of metrics to evaluate. Each metric should have a name attribute.

based_on_CI

Whether to compare each method with the separate analysis through the confidence intervals of the metric rather than its point estimates.

nominal_tie

The nominal type-I error threshold. Required when metrics includes success_proba.

Value

A dataframe containing the sweet spots for each metric, case study, method, and combination of parameters.

Examples

if (FALSE) { # \dontrun{
results <- sweet_spot(results_freq_df, metrics, nominal_tie = 0.025)
} # }