Reproducibility check

Mechanism

teal.data classes that accumulate raw data accept a boolean check argument, e.g. [teal_data()]. This boolean flag indicates whether teal performs validation of the data passed to the TealData object during launch of a teal application.

The validation compares the raw data passed to the class with the dataset returned by the code stored in the class. E.g.:

library(teal.data)
iris_ds <- dataset("iris", iris, code = "iris <- iris")
data <- teal_data(iris_ds, check = TRUE)
data$check() # returns TRUE

The check() method called during launch can be invoked directly and returns TRUE if the raw data passed to teal_data matches the return value of the code stored in the class. It returns an error if the comparison fails and NULL if the check = FALSE was passed to teal_data.

# throws an error:
# Error in x$check_reproducibility() : Reproducibility check failed.
data <- teal_data(dataset("iris", iris, code = "iris <- mtcars"), check = TRUE)

data <- teal_data(dataset("iris", iris, code = "iris <- mtcars"), check = FALSE)
data$check() # returns NULL even though the code does not match the raw data

Find out more in the teal_data() documentation.

Final considerations

<x>_connector objects do not accept raw data as input and therefore are not checked even if check = TRUE is used.

Reproducibility check

NEST CoreDev

2022-04-20