Execute simulations defined in a optic_simulation object

Usage

dispatch_simulations(object, seed = NULL, use_future = FALSE, verbose = 0, ...)

Arguments

object: Simulation scenarios object created using optic_simulation
seed: Specified as either NULL or a numeric. Sets a seed, which is becomes an index in results, for each independent set of simulations in optic_simulation.
use_future: Runs simulation scenarios in parallel. Default FALSE, set to TRUE if you have already setup a future plan (e.g., multiprocess, cluster, etc) and would like for the iterations to be run in parallel.
verbose: Default TRUE. IF TRUE, provides details on what's currently running.
...: additional parameters to be passed to future_apply. User can pass future.globals and future.packages if your code relies on additional packages

Value

A list of dataframes, where each list entry contains results for a set of simulation parameters, with dataframes containing estimated treatment effects and summary statistics by model and draw.

Examples

# Set up a basic model and simulation scenario:
data(overdoses)

eff <- 0.1*mean(overdoses$crude.rate, na.rm = TRUE)
form <- formula(crude.rate ~ state + year + population + treatment_level)
mod <- optic_model(name = 'lin', 
                   type = 'reg', 
                   call = 'lm', 
                   formula = form, 
                   se_adjust = 'none')

sim <- optic_simulation(x = overdoses, 
                        models = list(mod), 
                        method = 'no_confounding', 
                        unit_var = 'state', 
                        treat_var = 'state',
                        time_var = 'year', 
                        effect_magnitude = list(eff), 
                        n_units = 2, 
                        effect_direction = 'pos', 
                        iters = 2,
                        policy_speed = 'instant', 
                        n_implementation_periods = 1)
#> Number of Simulations: 1
#> Number of Models: 1
#> Iteration per Simulation : 2
#> Total number of Iterations to Run: 2

# Finally, dispatch the simulation:
dispatch_simulations(sim)
#> [[1]]
#>      outcome se_adjustment  estimate       se variance    t_stat     p_value
#> 1 crude.rate          none -1.298367 1.239787 1.537071 -1.047250 0.295263492
#> 2 crude.rate          none -3.775589 1.302310 1.696011 -2.899148 0.003832122
#>        mse model_name model_call
#> 1 10.98252        lin         lm
#> 2 10.85485        lin         lm
#>                                              model_formula policy_speed
#> 1 crude.rate ~ state + year + population + treatment_level      instant
#> 2 crude.rate ~ state + year + population + treatment_level      instant
#>   n_implementation_periods prior_control effect_magnitude n_units
#> 1                        1         level         1.265073       2
#> 2                        1         level         1.265073       2
#>   effect_direction n mean_es_prior max_es_prior mean_es_outcome max_es_outcome
#> 1              pos 2    -0.3928205    0.5397018     -0.07062395      0.6169953
#> 2              pos 2    -0.4702902    0.8307804     -0.47402847      0.8230322
#>   n_unique_enact_years mu1_prior mu0_prior sd_prior mu1_prior_old mu0_prior_old
#> 1                    2  11.93194  12.04081 5.098109      11.93194      12.04081
#> 2                    2  12.61818  12.02951 5.098109      12.61818      12.02951
#>   sd_prior_old      mu1      mu0       sd iter
#> 1     5.098109 14.24479 12.64049 6.651443    1
#> 2     5.098109 14.02575 12.64879 6.647637    2
#>