The bootstrap error
You work as an R programmer for a clinical laboratory. The statistician at the lab is conducting a large scale analysis to establish ranges for the average weights of different species of mice. She has written a a function, bootstrap()
, that is applied to every element of weight_list
in parallel using a futures backend. However the code throws the following error:
Error in checkForRemoteErrors(val) :
one node produced an error: missing values and NaN's not allowed if 'na.rm' is FALSE
She cannot find the element of weight_list
which causes the error. She also loses all results of the time-consuming calculation. You have been asked to fix the code. The furrr
package has been loaded for you and the multisession
for the futures has been planned.
This exercise is part of the course
Parallel Programming in R
Exercise instructions
- Supply the code in the curly braces to a function that can catch the error.
- To the error argument, supply a function that takes one argument,
e
, and returns the string"Error here!"
. - Map
bootstrap()
to all elements of theweight_list
using afurrr
function.
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
bootstrap <- function (x) {
# Supply code to a function for catching errors
___({est <- rep(0, 1e3)
for (i in 1:1e3) {
b <- sample(x, replace = T)
est[i] <- mean(b)
}
return(quantile(est, c(0.25, 0.75)))
# Supply a function that returns a string
}, error = ___
)
}
# Map bootstrap() to every element of weight_list
___(___, ___)
plan(sequential)