Anomaly Detection via H2O Deep Learning Model

Detect anomalies in an H2O dataset using an H2O deep learning model with auto-encoding.

h2o.anomaly(object, data, per_feature = FALSE)

Arguments

object	An H2OAutoEncoderModel object that represents the model to be used for anomaly detection.
data	An H2OFrame object.
per_feature	Whether to return the per-feature squared reconstruction error

Value

Returns an H2OFrame object containing the reconstruction MSE or the per-feature squared error.

See also

h2o.deeplearning for making an H2OAutoEncoderModel.

Examples

# NOT RUN {
library(h2o)
h2o.init()
prosPath = system.file("extdata", "prostate.csv", package = "h2o")
prostate.hex = h2o.importFile(path = prosPath)
prostate.dl = h2o.deeplearning(x = 3:9, training_frame = prostate.hex, autoencoder = TRUE,
                               hidden = c(10, 10), epochs = 5)
prostate.anon = h2o.anomaly(prostate.dl, prostate.hex)
head(prostate.anon)
prostate.anon.per.feature = h2o.anomaly(prostate.dl, prostate.hex, per_feature=TRUE)
head(prostate.anon.per.feature)
# }