CalculateROAS: Calculate ROAS or mROAS.
In google/amss: Agreggate Marketing System Simulator

Description Usage Arguments Value

This functions takes the original budget settings and a counterfactual budget setting. It reports the expected ratio between the total difference in revenue over all time points and the total difference in media spend over all time points.

CalculateROAS(object, new.budget = NULL,
  media.names = object$params$media.names, budget.periods = NULL,
  budget.proportion = rep(0, length(media.names)), t.start = 1,
  t.end = object$params$time.n, scaled.pop.size = 1e+18, min.reps = 2,
  max.coef.var = 0.01, max.margin.error = 0.01, max.time = 0,
  verbose = FALSE)

`object`	amss.sim object containing simulated data
`new.budget`	table of new budgets for each budget period (row) and media channel (column)
`media.names`	if new.budget is NULL, adjust original budget of the media named here.
`budget.periods`	budget.periods over which to modify the budget. Default `NULL` will lead to all budget periods being modified.
`budget.proportion`	nonnegative numeric. When `new.budget` is NULL, it is calculated by setting the budget of the media channels specified in `media.names` to `budget.proportion` proportion of the original budget during the budget periods specified in `budget.periods`. The default proportion of 0 is used to calculate the average ROAS over the entire spend in the channel. Values such as 0.99 can be used to calculate the marginal ROAS.
`t.start`	time point to start generating data according to the new settings.
`t.end`	last time point to generate data according to the new settings. In scenarios with lag, this should extend past the last time point in the modified budget periods in order to include lagged effects in the calculation.
`scaled.pop.size`	`CalculateROAS` scales up the population size to reduce the variability of its estimates. This is equivalent to running the simulation for multiple repetitions to reduce variability. The default value should provide sufficient accuracy in most use cases. Extremely large values may result in numerical issues.
`min.reps`	integer representing the initial number of datasets to generate from each budget setting. The default value of 2 allows the user to make a rough check that the accuracy is indeed good under the chosen settings. This default was chosen under the assumption that the default `scaled.pop.size` is large enough to accurately measure ROAS using 1 repetition, with the 2nd being used as confirmation of the accuracy. Higher precision and more accurate measurement of the precision can be achieved with more repetitions.
`max.coef.var`	numeric, the target coefficient of variation. The function takes additional samples of the ROAS until it runs out of time, attains the target coefficient of variation, or attains the target margin of error.
`max.margin.error`	numeric, the target margin of error. The function takes additional samples of the ROAS until it runs out of time, attains the target coefficient of variation, or attains the target margin of error.
`max.time`	numeric, the number of minutes at which to cut off the function from taking additional samples beyond the initial sample generated according to `min.reps`. The function takes additional samples of the ROAS until it runs out of time, attains the target coefficient of variation, or attains the target margin of error. The default value of 0 forces the function to use precisely `min.reps` repetitions.
`verbose`	boolean. If TRUE, output measures of the accuracy of the reported ROAS, including the full sample of ROAS values.