Add multi test into simfix

[LittleBeannie]

# Define different WLR tests into a list
tests <- list(
  test1 = tibble(rho = 0, gamma = 0),
  test2 = tibble(rho = 0, gamma = 0.5)
)

# Define a function to simulate one test method
sim1test <- function(x) {
  ans <- simfix(
    nsim = 3, sampleSize = 500, targetEvents = 350,
    strata = tibble(Stratum = "All", p = 1),
    enrollRates = tibble(duration = c(2, 2, 10), rate = c(3, 6, 9)),
    failRates = tibble(Stratum = "All", duration = c(3, 100), failRate = log(2) / c(9, 18), hr = c(.9, .6), dropoutRate = rep(.001, 2)),
    totalDuration = 30,
    block = rep(c("Experimental", "Control"), 2),
    timingType = 1:5,
    rg = x
  )
  return(ans)
}

# Run over different test methods
lapply(tests, FUN = sim1test)

[elong0527]

I thought it is more general to provide a list of functions with standardized input and output.

Use case; what if you want to perform simulation to compare WLR, RMST and milestone test.

In this way, you also have a framework to allow user defined function.

[nanxstats]

Perhaps something like this?

simfix(
  sampleSize = 500,
  targetEvents = 350,
  strata = strata_df(...),
  enrollRates = enroll_rates(...),
  failRates = fail_rates(...),
  totalDuration = 30,
  block = rep(c("Experimental", "Control"), 2),
  timingType = 1:5
) |>
  test_name_1() |>
  test_name_2(arg1 = 0, arg2 = 0)|>
  test_name_2(arg1 = 0, arg2 = 0.5) |>
  run_sim(n = 3)

[LittleBeannie]

Decision: let's focus on the data table first, and save this issue as low priority for now. The test list should be independent of rg. Better to have function calls for different tests.

[nanxstats]

Now after talking about this today I think Yilong's suggestion makes sense. For defining the tests, a function factory pattern like the ggplot2 scales or encapsulated OOP like R6 might be good solutions. We do need to change the arguments from the simple rg.