simulate_coverage {loewesadditivity}

Simulate a GIA model with an assumed error structure

Description

Simulate a GIA model with an assumed error structure

Usage

simulate_coverage(
  n_sims = 10,
  n_boot = 100,
  verbose = TRUE,
  experimental_grid,
  model_par,
  alpha = 0.05,
  noise_par = c(a0 = 2, a1 = 0.01),
  GIA_fn = base_GIA,
  S_fn = calc_S_base,
  fn_list = NULL
)

Arguments

n_sims

number of coverage simulations

n_boot

number of bootstraps to use in each simulation

verbose

logical indicating whether we should use print statements. Default is TRUE

experimental_grid

data frame with columns 'dose_A' and 'dose_B'

model_par

named vector of parameters corresponding to those used in GIA_fn()

alpha

alpha level used to produce confidence intervals for each bootstrap

noise_par

named vector for the noise parameter. Must have names "a0" and "a1". See ?base_gia for more details.

GIA_fn

function used to calculate GIA. Default is base_GIA().

S_fn

function to calculate S

fn_list

additional parameters to pass to GIA_fn

Value

list with the following entries

interaction_cov

This is the percent of times 0 was in the (1-alpha)% confidence interval for the interaction term "tau_1" from the simulated results

params_cov

This is the percent of times the true model parameter (those from model_par) lies in the (marginal) 95% confidence interval for that model parameter.

tau_pos

This is the percent of times the (1-alpha)% CI of "tau_1" was completely above 0.

tau_neg

This is the percent of times (1-alpha)% CI of "tau_1" is completely below zero

Examples

df <- loewesadditivity::cyrpa_ripr
df$dose_A <- df$CyRPA
df$dose_B <- df$RIPR
data <- fortify_gia_data(df)
model_params <- c("beta_A" = .247, "beta_B" = .224,
                  "gamma_A" = .734, "gamma_B" = .806,
                  "tau_1" = .28, "tau_2" = -.28)
experimental_grid <- make_grid(par = model_params,
                               n = 5)
n_boot <- 100
n_sims <- 10
GIA_fn <- base_GIA
S_fn <- calc_S_base
fn_list <- NULL
alpha <- .05
verbose <- TRUE
## NOT RUN
##out <- simulate_coverage(n_sims = n_sims,
  ##                      n_boot = n_boot,
   ##                     verbose = TRUE,
    ##                    experimental_grid = experimental_grid,
     ##                   model_par = model_params,
     ##                   alpha = .05,
   ##                     noise_par = c("a0" = 3, "a1" = .01),
   ##                     GIA_fn = base_GIA,
   ##                     fn_list = NULL)
##out