validate: Separate Validation of Trained Decision Forest

In Rborist: Extensible, Parallelizable Implementation of the Random Forest Algorithm

Separate Validation of Trained Decision Forest

Description

Permits trained decision forest to be validated separately from training.

Usage

## Default S3 method:
validate(train, preFormat, sampler = NULL,  ctgCensus
= "votes", impPermute = 0, quantVec = numeric(0), quantiles =
length(quantVec) > 0, indexing = FALSE, trapUnobserved = FALSE, nThread = 0, verbose =
FALSE, ...)

Arguments

train	an object of class Rborist obtained from previous training.
sampler	summarizes the response and its per-tree samplgin.
preFormat	internal representation of the design matrix, of class PreFormat
ctgCensus	report categorical validation by vote or by probability.
impPermute	specifies the number of importance permutations: 0 or 1.
quantVec	quantile levels to validate.
quantiles	whether to report quantiles at validation.
indexing	whether to report final index, typically terminal, of tree traversal.
trapUnobserved	indicates whether to return a nonterminal for values unobserved during training, such as missing data.
nThread	suggests an OpenMP-style thread count. Zero denotes the default processor setting.
verbose	indicates whether to output progress of validation.
...	not currently used.

Value

either of two pairs of objects:

• SummaryReg summarizing regression, as documented with the command predict.arbTrain.

• validation an object of class ValidReg consisting of:

  • mse the mean-square error of the estimate.

  • rsq the r-squared statistic of the estimate.

  • mae the mean absolute error of the estimate.

• SummaryCtg summarizing classification, as documented with the command predict.arbTrain.

• validation an object of class ValidCtg consisting of:

  • confusion the confusion matrix.

  • misprediction the misprediction rate.

  • oobError the out-of-bag error.

Author(s)

Mark Seligman at Suiji.

Examples

## Not run: 
    ## Trains without validation.
    rb <- Rborist(x, y, novalidate=TRUE)
    ...
    ## Delayed validation using a preformatted object.
    pf <- preformat(x)
    v <- validate(rb, pf)
  
## End(Not run)

Rborist documentation built on April 3, 2025, 8:04 p.m.