predict.iForest: predict.iForest

In Zelazny7/isofor: Isolation Forest Anomaly Detection

Description

return predictions of various types for the isolation forest object

Usage

## S3 method for class 'iForest'
predict(object, newdata, ..., nodes = FALSE,
  sparse = FALSE, replace_missing = TRUE, sentinel = -9999999999)

Arguments

object	an iForest object
newdata	a data.frame to predict
...	optional arguments not used.
nodes	if true return nobs x ntrees dim matrix with terminal node ids
sparse	if true return sparse Matrix of dimension nobs x nTerminalNodes. Each column represents a terminal node. There are as many ones in each row as there are trees in the forest. Each observation can only belong to one terminal node per tree. Useful for further modeling or to identify predictive interactions.
replace_missing	if TRUE, replaces missing factor levels with "." and missing numeric values with the sentinel argument
sentinel	value to use as stand-in for missing numeric values

Details

By default the predict function returns an anomaly score. The anomaly score is a [0,1] scaled measure of isolation. Higher scores correspond to more isolated observations. If sparse or nodes are set to TRUE, a matrix of the requested type is returned.

Value

A numeric vector of length nrow(newdata) containing values between zero and one. Values closer to zero are less likely to be anomalous.

Examples

mod <- iForest(iris, phi=16, nt=5)
score <- predict(mod, newdata = iris)

Zelazny7/isofor documentation built on Aug. 28, 2019, 7:12 p.m.