defaultOOBPerformanceAnalysis: Perform generic out-of-bag error analysis.

In boostr: A modular framework to bag or boost any estimation procedure.

Description

If performing regression, calculate which out-of-bag residuals and MSE. Otherwise, calculate which out-of-bag observations were classified correctly, what the overall misclassification rate is, as well as the confusion matrix.

Usage

defaultOOBPerformanceAnalysis(prediction, response, oobObs)

Arguments

prediction	a vector of predicted responses.
response	a vector of true response.
oobObs	a vector of indices which values in predictions are of out-of-bag observations.

Value

If performing regression, return a list with components:

oobMSE	the out-of-bag mean squared error.
resVec	a vector of length nrow(data) whose entries correspond to observations in data. The entry has values NA if the observation was not out-of-bag, and the difference between the predicted and true response (the residual) if the observation was out-of-bag.

Otherwise, return a list with components:

oobErr	overall misclassification rate.
oobConfMat	the confusion matrix of out-of-bag predictions against the true class labels.
errVec	a vector of length nrow(data) whose entries correspond to observations in data. The entry has values NA if the observation was not out-of-bag, and a 1 or 0 depending whether estimator failed to correctly classify the observation.

See Also

Other performance analyzers: boost, boost.function, boost.list

boostr documentation built on May 2, 2019, 1:42 p.m.