inside {multiverse}

Pass code into the multiverse

Description

Add code to the multiverse using the function using a function call, or an assignment operator, which is a wrapper around the function

Usage

inside(multiverse, .expr, .label = NULL, .execute_default = TRUE)

Arguments

multiverse

A multiverse object. A multiverse object is an S3 object which can be defined using multiverse()

.expr

R syntax. All the operations that the user wants to perform within the multiverse can be passed. Since it accepts a single argument, chunks of code can be passed using '{}'. See example for details.

.label

It is extracted automatically from the code block of type multiverse when run in an RMarkdown document. This should be used only within an RMarkdown document. Defaults to NULL.

.execute_default

Should the default multiverse be executed as part of this call?

Details

The inside function can only access variables which can be accessed at the same environment where the multiverse object was declared in.

To perform a multiverse analysis, we will need to write code to be executed within the multiverse. The inside() functions allows us to do this. Use inside() to pass any code to the specified multiverse, which is captured as an expression. To define multiple analysis options in the code passed to the multiverse, use the branch() function.

Value

a multiverse object

See Also

branch() for more details on how to declare multiple analysis options.

The inside function only stores the code, and does not execute any code at this step. To execute, we provide separate functions. See execute() for executing the code.

Instead of using the inside() function, an alternate implementation of the multiverse is using the assignment operator, '<-' (please refer to examples below).

Note: the inside() function can only access variables which can be accessed at the same level as the multiverse object. Since inside() is merely an interface to add analysis to the multiverse object, even if it is being called by another function, it is actually manipulating the multiverse object, which will have a different parent environment from where inside() is called, and hence not have access to variables which might be accessible in the environment within the function from where inside() is called.

Examples

M.1 <- multiverse()

# using `inside` to declare multiverse code
inside(M.1, {
  data <- rnorm(100, 50, 20)

  x.mean <- mean(data, trim = branch(
    trim_values,
    "trim_none" ~ 0,
    "trim_1pc" ~ 0.05,
    "trim_5pc" ~ 0.025,
    "trim_10pc" ~ 0.05
  ))
})

M.2 <- multiverse()

# using the assignment operator to declare multiverse code
inside(M.2, {
    data <- rnorm(100, 50, 20)
}) 

inside(M.2, {
    mean <- mean(data, trim = branch(
    trim_values,
    "trim_none" ~ 0,
    "trim_1pc" ~ 0.05,
    "trim_5pc" ~ 0.025,
    "trim_10pc" ~ 0.05
  ))
})

# declaring multiple options for a data processing step (calculating a new variable)
data(durante)
df <- durante

inside(M.1, {
  df <- df  %>%
    mutate( ComputedCycleLength = StartDateofLastPeriod - StartDateofPeriodBeforeLast ) %>%
    mutate( NextMenstrualOnset = branch(menstrual_calculation,
                                   "mc_option1" ~ StartDateofLastPeriod + ComputedCycleLength,
                                   "mc_option2" ~ StartDateofLastPeriod + ReportedCycleLength,
                                   "mc_option3" ~ StartDateNext
  ))
})