Share via


as.randomForest: Conversion of an rxDForest, rxDTree, or rpart object to an randomForest Object

Description

Converts objects containing decision tree results to an randomForest object.

Usage

 ## S3 method for class `rxDForest':
as.randomForest  (x, ...)
 ## S3 method for class `rxDTree':
as.randomForest  (x, ...)
 ## S3 method for class `rpart':
as.randomForest  (x, use.weight = TRUE, ties.method = c("random", "first", "last"), ...)

Arguments

x

object of class rxDForest, rxDTree, or rpart.

use.weight

a logical value (default being TRUE) specifying if the majority splitting direction at a node should be decided based on the sum of case weights or the number of observations when the split variable at the node is a factor or ordered factor but a certain level is not present (or not defined for the factor).

ties.method

a character string specifying how ties are handled when deciding the majority direction, with the default being "random". Refer to max.col for details.

...

additional arguments to be passed directly to as.randomForest.rpart.

Details

These functions convert an existing object of class rxDForest, rxDTree, or rpart to an object of class randomForest, respectively. The underlying structure of the output object will be a subset of that produced by an equivalent call to randomForest. Often, this method can be used to coerce an object for use with the pmml package. RevoScaleR model objects that contain transforms or a transformFunc are not supported.

Value

an object of class randomForest.

Author(s)

Microsoft Corporation Microsoft Technical Support

See Also

rxDForest, rxDTree, rpart, randomForest, as.rpart.

Examples


 ## Not run:

# If the pmml and the randomForest packages are installed 
library(pmml)
library(randomForest)

mydata <- infert
form <- case ~ age + parity + education + spontaneous + induced
ntree <- 2

fit.dforest <- rxDForest(form, data = mydata, nTree = ntree)    #, cp = 0.01, seed = 1
fit.dforest
fit.dforest.rforest <- as.randomForest(fit.dforest)
predict(fit.dforest.rforest, newdata = mydata)
pmml(fit.dforest.rforest)

fit.rforest <- randomForest(form, data = mydata, ntree = ntree)
fit.rforest
predict(fit.rforest, newdata = mydata)
pmml(fit.rforest)
## End(Not run)