run_cp_bootstrap: Run multiple bootstrapped change_point simulations (in...

In aarmiller/delaySim: Package to run delayed diagnosis simulations

Run multiple bootstrapped change_point simulations (in parallel)

Description

Run multiple bootstrapped change_point simulations (in parallel)

Usage

run_cp_bootstrap(
  sim_data,
  boot_trials = 100,
  n_sim_trials = 50,
  sim_algorithm = "simple",
  sim_ctrl = NULL,
  num_cores = NULL,
  no_bootstrapping = FALSE
)

Arguments

sim_data	A dataset containing the time_map, miss bins and other parameters used for the simulation. This dataset should be created using the 'prep_sim_data()' function
boot_trials	The number of bootstrapped trials to run (default is 100)
n_sim_trials	The number of trials to run in simulation of miss visits (default is 50)
num_cores	The number of worker cores to use. If not specified will determined the number of cores based on the which ever is the smallest value between number of boot_trials or detected number of cores - 1
no_bootstrapping	Specifies whether you want to run the simulations without bootstrapping the original dataset
new_draw_weight	A weighing parameter used to assign preference to drawing previously "missed" patients at each time step. A value of 0 applies strict preference to drawing patients who have been assigned to miss in prior time steps, while a value 0.5 applies equal weight to patients who have and have not been previously selected

Examples

### Run simulations with bootstrapping and allow change point to vary with each bootstrap sample ###

#load example final_time_map dataset
load("/Shared/Statepi_Diagnosis/grant_projects/hsv_enceph/scripts/validation/enrolled_ge_365/report_data.RData")

# rename ED column
final_time_map <- final_time_map %>% rename(ed = ED)

# run prep sim function
tmp_sim_data <- prep_sim_data(final_time_map, event_name = "any_ssd", cp_method = "cusum", start_day = 1L, by_days = 1L,
                              week_period = TRUE)

#run simulations on number of visits
simulation_results <- run_cp_bootstrap(tmp_sim_data,
                                       boot_trials = 500,
                                       n_sim_trials = 50,
                                       sim_algorithm = sim_algorithm,
                                       sim_ctrl = sim_ctrl,
                                       num_cores = NULL,
                                       no_bootstrapping = FALSE)


### Run simulations with bootstrapping and specify a change point applied to each bootstrap sample ###

# set a change point
cp <- 20L

# run prep sim function
tmp_sim_data <- prep_sim_data(final_time_map, event_name = "any_ssd", cp_method = "cusum", start_day = 1L, by_days = 1L,
                              week_period = TRUE, specify_cp = cp)

#run simulations on number of visits
simulation_results <- run_cp_bootstrap(tmp_sim_data,
                                       boot_trials = 500,
                                       n_sim_trials = 50,
                                       sim_algorithm = sim_algorithm,
                                       sim_ctrl = sim_ctrl,
                                       num_cores = NULL,
                                       no_bootstrapping = FALSE)


### Run simulations without bootstrapping and allow function to find the optimal change point for inputted data ###

# run prep sim function
tmp_sim_data <- prep_sim_data(final_time_map, event_name = "any_ssd", cp_method = "cusum", start_day = 1L, by_days = 1L,
                              week_period = TRUE)

#run simulations on number of visits
simulation_results <- run_cp_bootstrap(tmp_sim_data,
                                       boot_trials = 0,
                                       n_sim_trials = 25000,
                                       sim_algorithm = sim_algorithm,
                                       sim_ctrl = sim_ctrl,
                                       num_cores = NULL,
                                       no_bootstrapping = TRUE)


### Run simulations without bootstrapping and specify a change point instead of allowing function to find the optimal change point to apply to the data ###

# set a change point
cp <- 20L

# run prep sim function
tmp_sim_data <- prep_sim_data(final_time_map, event_name = "any_ssd", cp_method = "cusum", start_day = 1L, by_days = 1L,
                              week_period = TRUE, specify_cp = cp)

#run simulations on number of visits
simulation_results <- run_cp_bootstrap(tmp_sim_data,
                                       boot_trials = 0,
                                       n_sim_trials = 25000,
                                       sim_algorithm = sim_algorithm,
                                       sim_ctrl = sim_ctrl,
                                       num_cores = NULL,
                                       no_bootstrapping = TRUE)

aarmiller/delaySim documentation built on Jan. 2, 2023, 11:23 p.m.