Morris-style sensitivity screening from simulation replicates

Uses per-replicate variation to screen the most influential input parameters on an outcome metric. The implementation is a simple variance-based screening; for a full Morris design use the sensitivity package directly and pass the results here for visualisation.

Usage

morris_screening(data, outcome = "event_rate", inputs)

Arguments

data

A dynasimR_data object or summary tibble.

outcome

Character. Column name of the outcome to screen. Default "event_rate".

inputs

Character vector. Column names of simulation inputs (must be numeric) to assess.

Value

A tibble with columns input, mu_star (mean absolute change), sigma, rank.