Speed-accuracy trade-off
In the last video, you have seen there are two knobs you can tune to influence the performance of the random forests:
- Number of decision trees in each forest.
- Number of variables used for splitting within decision trees.
Increasing each of them might improve the accuracy of the imputation model, but it will also require more time to run. In this exercise, you will explore these ideas yourself by fitting missForest()
to the biopics
data twice with different settings. As you follow the instructions, pay attention to the errors you will be printing, and to the time the code takes to run.
This exercise is part of the course
Handling Missing Data with Imputations in R
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
# Set number of trees to 5 and number of variables used for splitting to 2
imp_res <- missForest(biopics, ___ = ___, ___ = ___)
# Print the resulting imputation errors
print(___)