MOA active learning classification

Description

MOA active learning classification

Usage

ActiveClassifier(control = NULL, ...)

Arguments

Value

An object of class MOA_classifier which sets up an untrained MOA model, which can be trained using trainMOA

See Also

MOAoptions, trainMOA

Examples

ctrl <- MOAoptions(model = "ActiveClassifier")
mymodel <- ActiveClassifier(control=ctrl)
mymodel

MOA bayesian classification

Description

MOA bayesian classification

Usage

NaiveBayes(control = NULL, ...)

NaiveBayesMultinomial(control = NULL, ...)

Arguments

Value

An object of class MOA_classifier which sets up an untrained MOA model, which can be trained using trainMOA

See Also

MOAoptions, trainMOA

Examples

ctrl <- MOAoptions(model = "NaiveBayes")
mymodel <- NaiveBayes(control=ctrl)
mymodel

MOA classification using ensembles

Description

MOA classification using ensembles (bagging/boosting/stacking/other)

Usage

AccuracyUpdatedEnsemble(control = NULL, ...)

AccuracyWeightedEnsemble(control = NULL, ...)

ADACC(control = NULL, ...)

DACC(control = NULL, ...)

LeveragingBag(control = NULL, ...)

LimAttClassifier(control = NULL, ...)

OCBoost(control = NULL, ...)

OnlineAccuracyUpdatedEnsemble(control = NULL, ...)

OzaBag(control = NULL, ...)

OzaBagAdwin(control = NULL, ...)

OzaBagASHT(control = NULL, ...)

OzaBoost(control = NULL, ...)

OzaBoostAdwin(control = NULL, ...)

TemporallyAugmentedClassifier(control = NULL, ...)

WeightedMajorityAlgorithm(control = NULL, ...)

Arguments

Value

An object of class MOA_classifier which sets up an untrained MOA model, which can be trained using trainMOA

See Also

MOAoptions, trainMOA

Examples

ctrl <- MOAoptions(model = "OzaBoostAdwin")
mymodel <- OzaBoostAdwin(control=ctrl)
mymodel

MOA classification trees

Description

MOA classification trees

Usage

AdaHoeffdingOptionTree(control = NULL, ...)

ASHoeffdingTree(control = NULL, ...)

DecisionStump(control = NULL, ...)

HoeffdingAdaptiveTree(control = NULL, ...)

HoeffdingOptionTree(control = NULL, ...)

HoeffdingTree(control = NULL, ...)

LimAttHoeffdingTree(control = NULL, ...)

RandomHoeffdingTree(control = NULL, ...)

Arguments

Value

An object of class MOA_classifier which sets up an untrained MOA model, which can be trained using trainMOA

See Also

MOAoptions, trainMOA

Examples

ctrl <- MOAoptions(model = "HoeffdingTree", leafprediction = "MC", 
   removePoorAtts = TRUE, binarySplits = TRUE, tieThreshold = 0.20)
hdt <- HoeffdingTree(control=ctrl)
hdt
hdt <- HoeffdingTree(numericEstimator = "GaussianNumericAttributeClassObserver")
hdt

Create a MOA classifier

Description

Create a MOA classifier

Usage

MOA_classifier(model, control = NULL, ...)

Arguments

Value

An object of class MOA_classifier

See Also

MOAoptions

Examples

RMOA:::.moaknownmodels
ctrl <- MOAoptions(model = "HoeffdingTree", leafprediction = "MC", 
   removePoorAtts = TRUE, binarySplits = TRUE, tieThreshold = 0.20)
hdt <- MOA_classifier(model = "HoeffdingTree", control=ctrl)
hdt
hdt <- MOA_classifier(
 model = "HoeffdingTree", 
 numericEstimator = "GaussianNumericAttributeClassObserver")
hdt

MOA recommendation engines

Description

MOA recommendation engines

Usage

BRISMFPredictor(control = NULL, ...)

BaselinePredictor(control = NULL, ...)

Arguments

Value

An object of class MOA_recommender which sets up an untrained MOA model, which can be trained using trainMOA

See Also

MOAoptions, trainMOA

Examples

ctrl <- MOAoptions(model = "BRISMFPredictor", features = 10)
brism <- BRISMFPredictor(control=ctrl)
brism
baseline <- BaselinePredictor()
baseline

Create a MOA recommendation engine

Description

Create a MOA recommendation engine

Usage

MOA_recommender(model, control = NULL, ...)

Arguments

Value

An object of class MOA_recommender

See Also

MOAoptions

Examples

RMOA:::.moaknownmodels
ctrl <- MOAoptions(model = "BRISMFPredictor", features = 10, lRate=0.002)
brism <- MOA_recommender(model = "BRISMFPredictor", control=ctrl)
brism
MOAoptions(model = "BaselinePredictor")
baseline <- MOA_recommender(model = "BaselinePredictor")
baseline

Create a MOA regressor

Description

Create a MOA regressor

Usage

MOA_regressor(model, control = NULL, ...)

Arguments

Value

An object of class MOA_regressor

See Also

MOAoptions

Examples

mymodel <- MOA_regressor(model = "FIMTDD")
mymodel
data(iris)
iris <- factorise(iris)
irisdatastream <- datastream_dataframe(data=iris)
## Train the model
mytrainedmodel <- trainMOA(model = mymodel, 
 Sepal.Length ~ Petal.Length + Species, data = irisdatastream)
mytrainedmodel$model

summary(lm(Sepal.Length ~ Petal.Length + Species, data = iris))
predict(mytrainedmodel, newdata=iris)

MOA regressors

Description

MOA regressors

Usage

TargetMean(control = NULL, ...)

FadingTargetMean(control = NULL, ...)

Perceptron(control = NULL, ...)

AMRulesRegressor(control = NULL, ...)

FIMTDD(control = NULL, ...)

ORTO(control = NULL, ...)

Arguments

Value

An object of class MOA_classifier which sets up an untrained MOA model, which can be trained using trainMOA

See Also

MOAoptions, trainMOA

Examples

ctrl <- MOAoptions(model = "FIMTDD", DoNotDetectChanges = TRUE, noAnomalyDetection=FALSE,
   univariateAnomalyprobabilityThreshold = 0.5, verbosity = 5)
mymodel <- FIMTDD(control=ctrl)
mymodel
mymodel <- FIMTDD(ctrlDoNotDetectChanges = FALSE)
mymodel

Define the attributes of a dataset (factor levels, numeric or string data) in a MOA setting

Description

Define the attributes of a dataset (factor levels, numeric or string data) in a MOA setting

Usage

MOAattributes(data, ...)

Arguments

Value

An object of class MOAmodelAttributes

Examples

data(iris)
mydata <- factorise(iris)
atts <- MOAattributes(data=mydata)
atts

Get and set options for models build with MOA.

Description

Get and set options for models build with MOA.

Usage

MOAoptions(model, ...)

Arguments

Value

An object of class MOAmodelOptions.
This is a list with elements:

1. model: The name of the model

2. moamodelname: The purpose of the model known by MOA (getPurposeString)

3. javaObj: a java reference of MOA options

4. options: a list with options of the MOA model. Each list element contains the Name of the option, the Purpose of the option and the current Value

See the examples.

Examples

control <- MOAoptions(model = "HoeffdingTree")
control
MOAoptions(model = "HoeffdingTree", leafprediction = "MC", 
   removePoorAtts = TRUE, binarySplits = TRUE, tieThreshold = 0.20)

## Other models known by RMOA
RMOA:::.moaknownmodels

## Classification Trees
MOAoptions(model = "AdaHoeffdingOptionTree")
MOAoptions(model = "ASHoeffdingTree")
MOAoptions(model = "DecisionStump")
MOAoptions(model = "HoeffdingAdaptiveTree")
MOAoptions(model = "HoeffdingOptionTree")
MOAoptions(model = "HoeffdingTree")
MOAoptions(model = "LimAttHoeffdingTree")
MOAoptions(model = "RandomHoeffdingTree")
## Classification using Bayes rule
MOAoptions(model = "NaiveBayes")
MOAoptions(model = "NaiveBayesMultinomial")
## Classification using Active learning
MOAoptions(model = "ActiveClassifier")
## Classification using Ensemble learning
MOAoptions(model = "AccuracyUpdatedEnsemble")
MOAoptions(model = "AccuracyWeightedEnsemble")
MOAoptions(model = "ADACC")
MOAoptions(model = "DACC")
MOAoptions(model = "LeveragingBag")
MOAoptions(model = "OCBoost")
MOAoptions(model = "OnlineAccuracyUpdatedEnsemble")
MOAoptions(model = "OzaBag")
MOAoptions(model = "OzaBagAdwin")
MOAoptions(model = "OzaBagASHT")
MOAoptions(model = "OzaBoost")
MOAoptions(model = "OzaBoostAdwin")
MOAoptions(model = "TemporallyAugmentedClassifier")
MOAoptions(model = "WeightedMajorityAlgorithm")

## Regressions
MOAoptions(model = "AMRulesRegressor")
MOAoptions(model = "FadingTargetMean")
MOAoptions(model = "FIMTDD")
MOAoptions(model = "ORTO")
MOAoptions(model = "Perceptron")
MOAoptions(model = "SGD")
MOAoptions(model = "TargetMean")

## Recommendation engines
MOAoptions(model = "BRISMFPredictor")
MOAoptions(model = "BaselinePredictor")

Datastream objects and methods

Description

Reference object of class datastream. This is a generic class which holds general information about the data stream.
Currently streams are implemented for data in table format (streams of read.table, read.csv, read.csv2, read.delim, read.delim2), data in RAM (data.frame, matrix), data in ff (on disk).
See the documentation of datastream_file, datastream_dataframe, datastream_matrix, and datastream_ffdf

Arguments

Value

A class of type datastream which contains

description:

character with the name how the stream is labelled.

state:

integer with the current state at which the stream will read new instances of data

processed:

integer with the number of instances already processed

finished:

logical indicating if the stream has finished processing all the instances

args:

list with arguments passed on to the stream when it is created (e.g. arguments of read.table)

See Also

datastream_file

Examples

## Basic example, showing the general methods available for a datastream object
x <- datastream(description = "My own datastream", args = list(a = "TEST"))
x
str(x)
try(x$get_points(x))

data streams on a data.frame

Description

Reference object of class datastream_dataframe. This is a class which inherits from class datastream and which can be used to read in a stream from a data.frame.

Arguments

Value

A class of type datastream_dataframe which contains

data:

The data.frame to extract instances from

all fields of the datastream superclass:

See datastream

Methods

• get_points(n) Get data from a datastream object.

  n

  integer, indicating the number of instances to retrieve from the datastream

See Also

datastream

Examples

x <- datastream_dataframe(data=iris)
x$get_points(10)
x
x$get_points(10)
x

data streams on an ffdf

Description

Reference object of class datastream_ffdf. This is a class which inherits from class datastream and which can be used to read in a stream from a ffdf from the ff package.

Arguments

Value

A class of type datastream_ffdf which contains

data:

The ffdf to extract instances from

all fields of the datastream superclass:

See datastream

Methods

• get_points(n) Get data from a datastream object.

  n

  integer, indicating the number of instances to retrieve from the datastream

See Also

datastream

Examples

## You need to load package ff before you can use datastream_ffdf
require(ff)
irisff <- as.ffdf(factorise(iris))
x <- datastream_ffdf(data=irisff)
x$get_points(10)
x
x$get_points(10)
x

File data stream

Description

Reference object of class datastream_file. This is a class which inherits from class datastream and which can be used to read in a stream from a file. A number of file readers have been implemented, namely datastream_table, datastream_csv, datastream_csv2, datastream_delim, datastream_delim2.
See the examples.

Arguments

Value

A class of type datastream_file which contains

FUN:

The function to use to read in the file

connection:

A connection to the file

columnnames:

A character vector of column names to overwrite the column names with in get_points

all fields of the datastream superclass:

See datastream

Methods

• get_points(n) Get data from a datastream object.

  n

  integer, indicating the number of instances to retrieve from the datastream

See Also

read.table, read.csv, read.csv2, read.delim, read.delim2

Examples

mydata <- iris
mydata$Species[2:3] <- NA
## Example of a CSV file stream
myfile <- tempfile()
write.csv(iris, file = myfile, row.names=FALSE, na = "")
x <- datastream_csv(file = myfile, na.strings = "")
x
x$get_points(n=10)
x
x$get_points(n=10)
x
x$stop()

## Create your own specific file stream
write.table(iris, file = myfile, row.names=FALSE, na = "")
x <- datastream_file(description="My file defintion stream", FUN=read.table, 
 file = myfile, header=TRUE, na.strings="")
x$get_points(n=10)
x
x$stop()


## Clean up for CRAN
file.remove(myfile)

data streams on a matrix

Description

Reference object of class datastream_matrix. This is a class which inherits from class datastream and which can be used to read in a stream from a matrix.

Arguments

Value

A class of type datastream_matrix which contains

data:

The matrix to extract instances from

all fields of the datastream superclass:

See datastream

Methods

• get_points(n) Get data from a datastream object.

  n

  integer, indicating the number of instances to retrieve from the datastream

See Also

datastream

Examples

data <- matrix(rnorm(1000*10), nrow = 1000, ncol = 10)
x <- datastream_matrix(data=data)
x$get_points(10)
x
x$get_points(10)
x

Convert character strings to factors in a dataset

Description

Convert character strings to factors in a dataset

Usage

factorise(x, ...)

Arguments

Value

a data.frame with the information in x where character columns are converted to factors

Examples

data(iris)
str(iris)
mydata <- factorise(iris)
str(mydata)

Predict using a MOA classifier, MOA regressor or MOA recommender on a new dataset

Description

Predict using a MOA classifier, MOA regressor or MOA recommender on a new dataset. \ Make sure the new dataset has the same structure and the same levels as get_points returns on the datastream which was used in trainMOA

Usage

## S3 method for class 'MOA_trainedmodel'
predict(object, newdata, type = "response",
  transFUN = object$transFUN, na.action = na.fail, ...)

Arguments

Value

A matrix of votes or a vector with the predicted class for MOA classifier or MOA regressor. A

See Also

trainMOA

Examples

## Hoeffdingtree
hdt <- HoeffdingTree(numericEstimator = "GaussianNumericAttributeClassObserver")
data(iris)
## Make a training set
iris <- factorise(iris)
traintest <- list()
traintest$trainidx <- sample(nrow(iris), size=nrow(iris)/2)
traintest$trainingset <- iris[traintest$trainidx, ]
traintest$testset <- iris[-traintest$trainidx, ]
irisdatastream <- datastream_dataframe(data=traintest$trainingset)
## Train the model
hdtreetrained <- trainMOA(model = hdt, 
 Species ~ Sepal.Length + Sepal.Width + Petal.Length + Petal.Width, 
 data = irisdatastream)

## Score the model on the holdoutset
scores <- predict(hdtreetrained, 
   newdata=traintest$testset[, c("Sepal.Length","Sepal.Width","Petal.Length","Petal.Width")], 
   type="response")
str(scores)
table(scores, traintest$testset$Species)
scores <- predict(hdtreetrained, newdata=traintest$testset, type="votes")
head(scores)

## Prediction based on recommendation engine
require(recommenderlab)
data(MovieLense)
x <- getData.frame(MovieLense)
x$itemid <- as.integer(as.factor(x$item))
x$userid <- as.integer(as.factor(x$user))
x$rating <- as.numeric(x$rating)
x <- head(x, 2000)

movielensestream <- datastream_dataframe(data=x)
movielensestream$get_points(3)

ctrl <- MOAoptions(model = "BRISMFPredictor", features = 10)
brism <- BRISMFPredictor(control=ctrl)
mymodel <- trainMOA(model = brism, rating ~ userid + itemid, 
 data = movielensestream, chunksize = 1000, trace=TRUE)

overview <- summary(mymodel$model)
str(overview)
predict(mymodel, head(x, 10), type = "response")

x <- expand.grid(userid=overview$users[1:10], itemid=overview$items)
predict(mymodel, x, type = "response")

Summary statistics of a MOA classifier

Description

Summary statistics of a MOA classifier

Usage

## S3 method for class 'MOA_classifier'
summary(object, ...)

Arguments

Value

the form of the return value depends on the type of MOA model

Examples

hdt <- HoeffdingTree(numericEstimator = "GaussianNumericAttributeClassObserver")
hdt
data(iris)
iris <- factorise(iris)
irisdatastream <- datastream_dataframe(data=iris)
## Train the model
hdtreetrained <- trainMOA(model = hdt, 
 Species ~ Sepal.Length + Sepal.Width + Petal.Length + Petal.Width, 
 data = irisdatastream)
summary(hdtreetrained$model)

Summary statistics of a MOA recommender

Description

Summary statistics of a MOA recommender

Usage

## S3 method for class 'MOA_recommender'
summary(object, ...)

Arguments

Value

the form of the return value depends on the type of MOA model

Examples

require(recommenderlab)
data(MovieLense)
x <- getData.frame(MovieLense)
x$itemid <- as.integer(as.factor(x$item))
x$userid <- as.integer(as.factor(x$user))
x$rating <- as.numeric(x$rating)
x <- head(x, 2000)

movielensestream <- datastream_dataframe(data=x)
movielensestream$get_points(3)

ctrl <- MOAoptions(model = "BRISMFPredictor", features = 10)
brism <- BRISMFPredictor(control=ctrl)
mymodel <- trainMOA(model = brism, rating ~ userid + itemid, 
 data = movielensestream, chunksize = 1000, trace=TRUE)

overview <- summary(mymodel$model)
str(overview)
predict(mymodel, head(x, 10), type = "response")

Summary statistics of a MOA regressor

Description

Summary statistics of a MOA regressor

Usage

## S3 method for class 'MOA_regressor'
summary(object, ...)

Arguments

Value

the form of the return value depends on the type of MOA model

Examples

## TODO

Train a MOA classifier/regressor/recommendation engine on a datastream

Description

Train a MOA classifier/regressor/recommendation engine on a datastream

Usage

trainMOA(model, ...)

Arguments

Value

An object of class MOA_trainedmodel which is returned by the methods for the specific model. See trainMOA.MOA_classifier, trainMOA.MOA_regressor, trainMOA.MOA_recommender

See Also

trainMOA.MOA_classifier, trainMOA.MOA_regressor, trainMOA.MOA_recommender

Train a MOA classifier (e.g. a HoeffdingTree) on a datastream

Description

Train a MOA classifier (e.g. a HoeffdingTree) on a datastream

Usage

## S3 method for class 'MOA_classifier'
trainMOA(model, formula, data, subset,
  na.action = na.exclude, transFUN = identity, chunksize = 1000,
  reset = TRUE, trace = FALSE, options = list(maxruntime = +Inf), ...)

Arguments

Value

An object of class MOA_trainedmodel which is a list with elements

• model: the updated supplied model object of class MOA_classifier

• call: the matched call

• na.action: the value of na.action

• terms: the terms in the model

• transFUN: the transFUN argument

See Also

MOA_classifier, datastream_file, datastream_dataframe, datastream_matrix, datastream_ffdf, datastream, predict.MOA_trainedmodel

Examples

hdt <- HoeffdingTree(numericEstimator = "GaussianNumericAttributeClassObserver")
hdt
data(iris)
iris <- factorise(iris)
irisdatastream <- datastream_dataframe(data=iris)
irisdatastream$get_points(3)

mymodel <- trainMOA(model = hdt, Species ~ Sepal.Length + Sepal.Width + Petal.Length, 
 data = irisdatastream, chunksize = 10)
mymodel$model
irisdatastream$reset()
mymodel <- trainMOA(model = hdt, 
 Species ~ Sepal.Length + Sepal.Width + Petal.Length + Petal.Length^2, 
 data = irisdatastream, chunksize = 10, reset=TRUE, trace=TRUE)
mymodel$model

Train a MOA recommender (e.g. a BRISMFPredictor) on a datastream

Description

Train a MOA recommender (e.g. a BRISMFPredictor) on a datastream

Usage

## S3 method for class 'MOA_recommender'
trainMOA(model, formula, data, subset,
  na.action = na.exclude, transFUN = identity, chunksize = 1000,
  trace = FALSE, options = list(maxruntime = +Inf), ...)

Arguments

Value

An object of class MOA_trainedmodel which is a list with elements

• model: the updated supplied model object of class MOA_recommender

• call: the matched call

• na.action: the value of na.action

• terms: the terms in the model

• transFUN: the transFUN argument

See Also

MOA_recommender, datastream_file, datastream_dataframe, datastream_matrix, datastream_ffdf, datastream, predict.MOA_trainedmodel

Examples

require(recommenderlab)
data(MovieLense)
x <- getData.frame(MovieLense)
x$itemid <- as.integer(as.factor(x$item))
x$userid <- as.integer(as.factor(x$user))
x$rating <- as.numeric(x$rating)
x <- head(x, 5000)

movielensestream <- datastream_dataframe(data=x)
movielensestream$get_points(3)

ctrl <- MOAoptions(model = "BRISMFPredictor", features = 10)
brism <- BRISMFPredictor(control=ctrl)
mymodel <- trainMOA(model = brism, rating ~ userid + itemid, 
 data = movielensestream, chunksize = 1000, trace=TRUE)
summary(mymodel$model)

Train a MOA regressor (e.g. a FIMTDD) on a datastream

Description

Train a MOA regressor (e.g. a FIMTDD) on a datastream

Usage

## S3 method for class 'MOA_regressor'
trainMOA(model, formula, data, subset,
  na.action = na.exclude, transFUN = identity, chunksize = 1000,
  reset = TRUE, trace = FALSE, options = list(maxruntime = +Inf), ...)

Arguments

Value

An object of class MOA_trainedmodel which is a list with elements

• model: the updated supplied model object of class MOA_regressor

• call: the matched call

• na.action: the value of na.action

• terms: the terms in the model

• transFUN: the transFUN argument

See Also

MOA_regressor, datastream_file, datastream_dataframe, datastream_matrix, datastream_ffdf, datastream, predict.MOA_trainedmodel

Examples

mymodel <- MOA_regressor(model = "FIMTDD")
mymodel
data(iris)
iris <- factorise(iris)
irisdatastream <- datastream_dataframe(data=iris)
irisdatastream$get_points(3)
## Train the model
mytrainedmodel <- trainMOA(model = mymodel, 
 Sepal.Length ~ Petal.Length + Species, data = irisdatastream)
mytrainedmodel$model
irisdatastream$reset()
mytrainedmodel <- trainMOA(model = mytrainedmodel$model, 
 Sepal.Length ~ Petal.Length + Species, data = irisdatastream, 
 chunksize = 10, reset=FALSE, trace=TRUE)
mytrainedmodel$model