This exercise will compare alternative measures of accuracy from randomForest() runs. First, 16 rows where data (on V6) is missing will be omitted: (a) Compare repeated randomForest() runs: (b)...

This exercise will compare alternative measures of accuracy from randomForest() runs. First, 16 rows where data (on V6) is missing will be omitted:

(a) Compare repeated randomForest() runs:

(b) Compare OOB accuracies with test set accuracies:

Plot test set accuracies against OOB accuracies. Add the line y = x to the plot. Is there any consistent difference in the accuracies? Given a random training/test split, is there any reason to expect a consistent difference between OOB accuracy and test accuracy?

(c) Calculate the error rate for the training data:

Explain why use of the training data for testing leads to an error rate that is zero.