Computes the Accuracy measure.

Description

Computes the ratio of number of correct predictions to the total number of input samples.

Details

Accuracy = (Number Correct Predictions) / (Total Number of Predictions)

Super class

D2MCS::MeasureFunction -> Accuracy

Methods

Public methods

• Accuracy$new()

• Accuracy$compute()

• Accuracy$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

Accuracy$new(performance.output = NULL)

Arguments

Method compute()

The function computes the Accuracy achieved by the M.L. model.

Usage

Accuracy$compute(performance.output = NULL)

Arguments

Details

This function is automatically invoke by the ClassificationOutput object.

Returns

A numeric vector of size 1 or NULL if an error occurred.

Method clone()

The objects of this class are cloneable with this method.

Usage

Accuracy$clone(deep = FALSE)

Arguments

See Also

MeasureFunction, ClassificationOutput, ConfMatrix.

Plotting feature clusters following bi-class problem.

Description

The BinaryPlot implements a basic plot for bi-class problem.

Super class

D2MCS::GenericPlot -> BinaryPlot

Methods

Public methods

• BinaryPlot$new()

• BinaryPlot$plot()

• BinaryPlot$clone()

Method new()

Empty function used to initialize the object arguments in runtime.

Usage

BinaryPlot$new()

Method plot()

Plots feature-clustering data from a bi-class problem.

Usage

BinaryPlot$plot(summary)

Arguments

Method clone()

The objects of this class are cloneable with this method.

Usage

BinaryPlot$clone(deep = FALSE)

Arguments

See Also

GenericPlot

Feature-clustering based on ChiSquare method.

Description

Performs feature-clustering based on ChiSquare method.

Super class

D2MCS::GenericHeuristic -> ChiSquareHeuristic

Methods

Public methods

• ChiSquareHeuristic$new()

• ChiSquareHeuristic$heuristic()

• ChiSquareHeuristic$clone()

Method new()

Empty function used to initialize the object arguments in runtime.

Usage

ChiSquareHeuristic$new()

Method heuristic()

Functions responsible of performing the ChiSquare feature-clustering operation.

Usage

ChiSquareHeuristic$heuristic(col1, col2, column.names = NULL)

Arguments

Returns

A numeric vector of length 1 or NA if an error occurs.

Method clone()

The objects of this class are cloneable with this method.

Usage

ChiSquareHeuristic$clone(deep = FALSE)

Arguments

See Also

Dataset, chisq.test

Implementation of Majority Voting voting.

Description

Implementation of the parliamentary 'majority voting' procedure. The majority class value is defined as final class. All class values have the same importance.

Super class

D2MCS::SimpleVoting -> ClassMajorityVoting

Methods

Public methods

• ClassMajorityVoting$new()

• ClassMajorityVoting$getMajorityClass()

• ClassMajorityVoting$getClassTie()

• ClassMajorityVoting$execute()

• ClassMajorityVoting$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

ClassMajorityVoting$new(cutoff = 0.5, class.tie = NULL, majority.class = NULL)

Arguments

Method getMajorityClass()

The function returns the value of the majority class.

Usage

ClassMajorityVoting$getMajorityClass()

Returns

A character vector of length 1 with the name of the majority class.

Method getClassTie()

The function gets the class value assigned to solve ties.

Usage

ClassMajorityVoting$getClassTie()

Returns

A character vector of length 1.

Method execute()

The function implements the majority voting procedure.

Usage

ClassMajorityVoting$execute(predictions, verbose = FALSE)

Arguments

Method clone()

The objects of this class are cloneable with this method.

Usage

ClassMajorityVoting$clone(deep = FALSE)

Arguments

See Also

D2MCS, ClassMajorityVoting, ClassWeightedVoting, ProbAverageVoting, ProbAverageWeightedVoting, ProbBasedMethodology

Implementation Weighted Voting scheme.

Description

A new implementation of ClassMajorityVoting where each class value has different values (weights).

Super class

D2MCS::SimpleVoting -> ClassWeightedVoting

Methods

Public methods

• ClassWeightedVoting$new()

• ClassWeightedVoting$getWeights()

• ClassWeightedVoting$setWeights()

• ClassWeightedVoting$execute()

• ClassWeightedVoting$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

ClassWeightedVoting$new(cutoff = 0.5, weights = NULL)

Arguments

Method getWeights()

The function returns the weights used to perform the voting scheme.

Usage

ClassWeightedVoting$getWeights()

Returns

A numeric vector.

Method setWeights()

The function allows changing the value of the weights.

Usage

ClassWeightedVoting$setWeights(weights)

Arguments

Method execute()

The function implements the cluster-weighted majority voting procedure.

Usage

ClassWeightedVoting$execute(predictions, verbose = FALSE)

Arguments

Method clone()

The objects of this class are cloneable with this method.

Usage

ClassWeightedVoting$clone(deep = FALSE)

Arguments

See Also

D2MCS, ClassMajorityVoting, ClassWeightedVoting, ProbAverageVoting, ProbAverageWeightedVoting, ProbBasedMethodology

D2MCS Classification Output.

Description

Allows computing the classification performance values achieved by D2MCS. The class is automatically created when D2MCS classification method is invoked.

Methods

Public methods

• ClassificationOutput$new()

• ClassificationOutput$getMetrics()

• ClassificationOutput$getPositiveClass()

• ClassificationOutput$getModelInfo()

• ClassificationOutput$getPerformances()

• ClassificationOutput$savePerformances()

• ClassificationOutput$plotPerformances()

• ClassificationOutput$getPredictions()

• ClassificationOutput$savePredictions()

• ClassificationOutput$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

ClassificationOutput$new(voting.schemes, models)

Arguments

Method getMetrics()

The function returns the measures used during training stage.

Usage

ClassificationOutput$getMetrics()

Returns

A character vector or NULL if training was not performed.

Method getPositiveClass()

The function gets the name of the positive class used for training/classification.

Usage

ClassificationOutput$getPositiveClass()

Returns

A character vector of size 1.

Method getModelInfo()

The function compiled all the information concerning to the M.L. models used during training/classification.

Usage

ClassificationOutput$getModelInfo(metrics = NULL)

Arguments

Returns

A list with the information of each M.L. model.

Method getPerformances()

The function is used to compute the performance of D2MCS.

Usage

ClassificationOutput$getPerformances(
  test.set,
  measures,
  voting.names = NULL,
  metric.names = NULL,
  cutoff.values = NULL
)

Arguments

Returns

A list of performance values.

Method savePerformances()

The function is used to save the computed predictions into a CSV file.

Usage

ClassificationOutput$savePerformances(
  dir.path,
  test.set,
  measures,
  voting.names = NULL,
  metric.names = NULL,
  cutoff.values = NULL
)

Arguments

Method plotPerformances()

The function allows to graphically visualize the computed performance.

Usage

ClassificationOutput$plotPerformances(
  dir.path,
  test.set,
  measures,
  voting.names = NULL,
  metric.names = NULL,
  cutoff.values = NULL
)

Arguments

Method getPredictions()

The function is used to obtain the computed predictions.

Usage

ClassificationOutput$getPredictions(
  voting.names = NULL,
  metric.names = NULL,
  cutoff.values = NULL,
  type = NULL,
  target = NULL,
  filter = FALSE
)

Arguments

Returns

A PredictionOutput object.

Method savePredictions()

The function saves the predictions into a CSV file.

Usage

ClassificationOutput$savePredictions(
  dir.path,
  voting.names = NULL,
  metric.names = NULL,
  cutoff.values = NULL,
  type = NULL,
  target = NULL,
  filter = FALSE
)

Arguments

Method clone()

The objects of this class are cloneable with this method.

Usage

ClassificationOutput$clone(deep = FALSE)

Arguments

See Also

D2MCS

Manages the predictions achieved on a cluster.

Description

Stores the predictions achieved by the best M.L. of each cluster.

Methods

Public methods

• ClusterPredictions$new()

• ClusterPredictions$add()

• ClusterPredictions$get()

• ClusterPredictions$getAll()

• ClusterPredictions$size()

• ClusterPredictions$getPositiveClass()

• ClusterPredictions$getClassValues()

• ClusterPredictions$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

ClusterPredictions$new(class.values, positive.class)

Arguments

Method add()

The function is used to add the prediction achieved by a specific M.L. model.

Usage

ClusterPredictions$add(prediction)

Arguments

Method get()

The function returns the predictions placed at specific position.

Usage

ClusterPredictions$get(position)

Arguments

Returns

A Prediction object.

Method getAll()

The function returns all the predictions.

Usage

ClusterPredictions$getAll()

Returns

A list containing all computed predictions.

Method size()

The function returns the number of computed predictions.

Usage

ClusterPredictions$size()

Returns

A numeric value.

Method getPositiveClass()

The function gets the value of the positive class.

Usage

ClusterPredictions$getPositiveClass()

Returns

A character vector of size 1.

Method getClassValues()

The function returns all the values of the target class.

Usage

ClusterPredictions$getClassValues()

Returns

A character vector containing all target values.

Method clone()

The objects of this class are cloneable with this method.

Usage

ClusterPredictions$clone(deep = FALSE)

Arguments

See Also

D2MCS, ClassificationOutput, Prediction

Abstract class to compute the class prediction based on combination between metrics.

Description

Abstract class used as a template to define new customized strategies to combine the class predictions made by different metrics.

Methods

Public methods

• CombinedMetrics$new()

• CombinedMetrics$getRequiredMetrics()

• CombinedMetrics$getFinalPrediction()

• CombinedMetrics$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

CombinedMetrics$new(required.metrics)

Arguments

Method getRequiredMetrics()

The function returns the required metrics that will participate in the combined metric process.

Usage

CombinedMetrics$getRequiredMetrics()

Returns

A character vector of length greater than 2 with the name of the required metrics.

Method getFinalPrediction()

Function used to implement the strategy to obtain the final prediction based on different metrics.

Usage

CombinedMetrics$getFinalPrediction(
  raw.pred,
  prob.pred,
  positive.class,
  negative.class
)

Arguments

Returns

A logical value indicating if the instance is predicted as positive class or not.

Method clone()

The objects of this class are cloneable with this method.

Usage

CombinedMetrics$clone(deep = FALSE)

Arguments

See Also

CombinedVoting

Implementation of Combined Voting.

Description

Calculates the final prediction by performing the result of the predictions of different metrics obtained through a SimpleVoting class.

Super class

D2MCS::VotingStrategy -> CombinedVoting

Methods

Public methods

• CombinedVoting$new()

• CombinedVoting$getCombinedMetrics()

• CombinedVoting$getMethodology()

• CombinedVoting$getFinalPred()

• CombinedVoting$execute()

• CombinedVoting$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

CombinedVoting$new(voting.schemes, combined.metrics, methodology, metrics)

Arguments

Method getCombinedMetrics()

The function returns the metrics used to combine the metrics results.

Usage

CombinedVoting$getCombinedMetrics()

Returns

An object inherited from CombinedMetrics class.

Method getMethodology()

The function gets the methodology used to execute the combined votings.

Usage

CombinedVoting$getMethodology()

Returns

An object inherited from Methodology class.

Method getFinalPred()

The function returns the predictions obtained after executing the combined-voting methodology.

Usage

CombinedVoting$getFinalPred(type = NULL, target = NULL, filter = NULL)

Arguments

Returns

A data.frame with the computed predictions.

Method execute()

The function implements the combined voting scheme.

Usage

CombinedVoting$execute(predictions, verbose = FALSE)

Arguments

Method clone()

The objects of this class are cloneable with this method.

Usage

CombinedVoting$clone(deep = FALSE)

Arguments

See Also

D2MCS, ClassMajorityVoting, ClassWeightedVoting, ProbAverageVoting, ProbAverageWeightedVoting, ProbBasedMethodology, SimpleVoting

Confusion matrix wrapper.

Description

Creates a R6 confusion matrix from the confusionMatrix caret package.

Methods

Public methods

• ConfMatrix$new()

• ConfMatrix$getConfusionMatrix()

• ConfMatrix$getTP()

• ConfMatrix$getTN()

• ConfMatrix$getFN()

• ConfMatrix$getFP()

• ConfMatrix$clone()

Method new()

Method to create a confusion matrix object from a caret confusionMatrix

Usage

ConfMatrix$new(confMatrix)

Arguments

Method getConfusionMatrix()

The function obtains the confusionMatrix following the same structured as defined in the caret package

Usage

ConfMatrix$getConfusionMatrix()

Returns

A confusionMatrix object.

Method getTP()

The function is used to compute the number of True Positive values achieved.

Usage

ConfMatrix$getTP()

Returns

A numeric vector of size 1.

Method getTN()

The function computes the True Negative values.

Usage

ConfMatrix$getTN()

Returns

A numeric vector of size 1.

Method getFN()

The function returns the number of Type II errors (False Negative).

Usage

ConfMatrix$getFN()

Returns

A numeric vector of size 1.

Method getFP()

The function returns the number of Type I errors (False Negative).

Usage

ConfMatrix$getFP()

Returns

A numeric vector of size 1.

Method clone()

The objects of this class are cloneable with this method.

Usage

ConfMatrix$clone(deep = FALSE)

Arguments

See Also

D2MCS, MeasureFunction, ClassificationOutput

Data Driven Multiple Classifier System.

Description

The class is responsible of managing the whole process. Concretely builds the M.L. models (optimizes models hyperparameters), selects the best M.L. model for each cluster and executes the classification stage.

Methods

Public methods

• D2MCS$new()

• D2MCS$train()

• D2MCS$classify()

• D2MCS$getAvailableModels()

• D2MCS$clone()

Method new()

The function is used to initialize all parameters needed to build a Multiple Classifier System.

Usage

D2MCS$new(
  dir.path,
  num.cores = NULL,
  socket.type = "PSOCK",
  outfile = NULL,
  serialize = FALSE
)

Arguments

Method train()

The function is responsible of performing the M.L. model training stage.

Usage

D2MCS$train(
  train.set,
  train.function,
  num.clusters = NULL,
  model.recipe = DefaultModelFit$new(),
  ex.classifiers = c(),
  ig.classifiers = c(),
  metrics = NULL,
  saveAllModels = FALSE
)

Arguments

Returns

A TrainOutput object containing all the information computed during the training stage.

Method classify()

The function is responsible for executing the classification stage.

Usage

D2MCS$classify(train.output, subset, voting.types, positive.class = NULL)

Arguments

Returns

A ClassificationOutput with all the values computed during classification stage.

Method getAvailableModels()

The function obtains all the available M.L. models.

Usage

D2MCS$getAvailableModels()

Returns

A data.frame containing the information of the available M.L. models.

Method clone()

The objects of this class are cloneable with this method.

Usage

D2MCS$clone(deep = FALSE)

Arguments

See Also

Dataset, Subset, Trainset

Examples

# Specify the random number generation
set.seed(1234)

## Create Dataset Handler object.
loader <- DatasetLoader$new()

## Load 'hcc-data-complete-balanced.csv' dataset file.
data <- loader$load(filepath = system.file(file.path("examples",
                                                     "hcc-data-complete-balanced.csv"),
                                           package = "D2MCS"),
                    header = TRUE, normalize.names = TRUE)
## Get column names
data$getColumnNames()

## Split data into 4 partitions keeping balance ratio of 'Class' column.
data$createPartitions(num.folds = 4, class.balance = "Class")

## Create a subset comprising the first 2 partitions for clustering purposes.
cluster.subset <- data$createSubset(num.folds = c(1, 2), class.index = "Class",
                                    positive.class = "1")

## Create a subset comprising second and third partitions for trainning purposes.
train.subset <- data$createSubset(num.folds = c(2, 3), class.index = "Class",
                                  positive.class = "1")

## Create a subset comprising last partitions for testing purposes.
test.subset <- data$createSubset(num.folds = 4, class.index = "Class",
                                 positive.class = "1")

## Distribute the features into clusters using MCC heuristic.
distribution <- SimpleStrategy$new(subset = cluster.subset,
                                   heuristic = MCCHeuristic$new())
distribution$execute()

## Get the best achieved distribution
distribution$getBestClusterDistribution()

## Create a train set from the computed clustering distribution
train.set <- distribution$createTrain(subset = train.subset)

## Not run: 

## Initialization of D2MCS configuration parameters.
##  - Defining training operation.
##    + 10-fold cross-validation
##    + Use only 1 CPU core.
##    + Seed was set to ensure straightforward reproductivity of experiments.
trFunction <- TwoClass$new(method = "cv", number = 10, savePredictions = "final",
                           classProbs = TRUE, allowParallel = TRUE,
                           verboseIter = FALSE, seed = 1234)

#' ## - Specify the models to be trained
ex.classifiers <- c("ranger", "lda", "lda2")

## Initialize D2MCS
#' d2mcs <- D2MCS$new(dir.path = tempdir(),
                      num.cores = 1)

## Execute training stage for using 'MCC' and 'PPV' measures to optimize model hyperparameters.
trained.models <- d2mcs$train(train.set = train.set,
                              train.function = trFunction,
                              ex.classifiers = ex.classifiers,
                              metrics = c("MCC", "PPV"))

## Execute classification stage using two different voting schemes
predictions <- d2mcs$classify(train.output = trained.models,
                              subset = test.subset,
                              voting.types = c(
                                    SingleVoting$new(voting.schemes = c(ClassMajorityVoting$new(),
                                                                        ClassWeightedVoting$new()),
                                                     metrics = c("MCC", "PPV"))))

## Compute the performance of each voting scheme using PPV and MMC measures.
predictions$getPerformances(test.subset, measures = list(MCC$new(), PPV$new()))

## Execute classification stage using multiple voting schemes (simple and combined)
predictions <- d2mcs$classify(train.output = trained.models,
                              subset = test.subset,
                              voting.types = c(
                                    SingleVoting$new(voting.schemes = c(ClassMajorityVoting$new(),
                                                                         ClassWeightedVoting$new()),
                                                      metrics = c("MCC", "PPV")),
                                    CombinedVoting$new(voting.schemes = ClassMajorityVoting$new(),
                                                        combined.metrics = MinimizeFP$new(),
                                                        methodology = ProbBasedMethodology$new(),
                                                        metrics = c("MCC", "PPV"))))

## Compute the performance of each voting scheme using PPV and MMC measures.
predictions$getPerformances(test.subset, measures = list(MCC$new(), PPV$new()))

## End(Not run)

Iterator over a Subset object

Description

Creates a DIterator object to iterate over the Subset.

Methods

Public methods

• DIterator$new()

• DIterator$getNext()

• DIterator$isLast()

• DIterator$finalize()

• DIterator$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

DIterator$new(data, chunk.size, verbose)

Arguments

Method getNext()

Gets the next chunk of data. Each iteration returns the same instances (data.frame rows) as chunk.size. However, if remaining data if less than chunk size, all the remaining data is returned. Conversely, NULL when there is no more pending data. By default chunk.size is defined as 10000.

Usage

DIterator$getNext()

Returns

A data.frame of NULL if all the data have been previously returned.

Method isLast()

Checks if the DIterator object reached the end of the data.frame

Usage

DIterator$isLast()

Returns

A logical value indicating if the end of data.frame has been reached.

Method finalize()

Destroys the DIterator object.

Usage

DIterator$finalize()

Method clone()

The objects of this class are cloneable with this method.

Usage

DIterator$clone(deep = FALSE)

Arguments

See Also

Dataset

Simple Dataset handler.

Description

Creates a valid simple dataset object.

Methods

Public methods

• Dataset$new()

• Dataset$getColumnNames()

• Dataset$getDataset()

• Dataset$getNcol()

• Dataset$getNrow()

• Dataset$getRemovedColumns()

• Dataset$cleanData()

• Dataset$removeColumns()

• Dataset$createPartitions()

• Dataset$createSubset()

• Dataset$createTrain()

Method new()

Method for initializing the object arguments during runtime.

Usage

Dataset$new(
  filepath,
  header = TRUE,
  sep = ",",
  skip = 0,
  normalize.names = FALSE,
  string.as.factor = FALSE,
  ignore.columns = NULL
)

Arguments

Method getColumnNames()

Get the name of the columns comprising the dataset.

Usage

Dataset$getColumnNames()

Returns

A character vector with the name of each column.

Method getDataset()

Gets the full dataset.

Usage

Dataset$getDataset()

Returns

A data.frame with all the loaded information.

Method getNcol()

Obtains the number of columns present in the dataset.

Usage

Dataset$getNcol()

Returns

An integer of length 1 or NULL

Method getNrow()

Obtains the number of rows present in the dataset.

Usage

Dataset$getNrow()

Returns

An integer of length 1 or NULL

Method getRemovedColumns()

Get the columns removed or ignored.

Usage

Dataset$getRemovedColumns()

Returns

A list containing the name of the removed columns.

Method cleanData()

Removes data.frame columns matching some criterion.

Usage

Dataset$cleanData(remove.funcs = NULL, remove.na = TRUE, remove.const = FALSE)

Arguments

Method removeColumns()

Applies cleanData function over an specific set of columns.

Usage

Dataset$removeColumns(
  columns,
  remove.funcs = NULL,
  remove.na = FALSE,
  remove.const = FALSE
)

Arguments

Method createPartitions()

Creates a k-folds partition from the initial dataset.

Usage

Dataset$createPartitions(
  num.folds = NULL,
  percent.folds = NULL,
  class.balance = NULL
)

Arguments

Method createSubset()

Creates a Subset for testing or classification purposes. A target class should be provided for testing purposes.

Usage

Dataset$createSubset(
  num.folds = NULL,
  opts = list(remove.na = TRUE, remove.const = FALSE),
  class.index = NULL,
  positive.class = NULL
)

Arguments

Returns

A Subset object.

Method createTrain()

Creates a set for training purposes. A class should be defined to guarantee full-compatibility with supervised models.

Usage

Dataset$createTrain(
  class.index,
  positive.class,
  num.folds = NULL,
  opts = list(remove.na = TRUE, remove.const = FALSE)
)

Arguments

Returns

A Trainset object.

See Also

HDDataset

Dataset creation.

Description

Wrapper class able to automatically create a Dataset, HDDataset according to the input data.

Methods

Public methods

• DatasetLoader$new()

• DatasetLoader$load()

Method new()

Empty function used to initialize the object arguments in runtime.

Usage

DatasetLoader$new()

Method load()

Stores the input source into a Dataset or HDDataset type object.

Usage

DatasetLoader$load(
  filepath,
  header = TRUE,
  sep = ",",
  skip.lines = 0,
  normalize.names = FALSE,
  string.as.factor = FALSE,
  ignore.columns = NULL
)

Arguments

Returns

A Dataset or HDDataset object.

See Also

Dataset, HDDataset

Examples

## Not run: 
# Create Dataset Handler object.
  loader <- DatasetLoader$new()

  # Load input file.
  data <- loader$load(filepath = system.file(file.path("examples",
                                                       "hcc-data-complete-balanced.csv"),
                                             package = "D2MCS"),
                      header = T, normalize.names = T)

## End(Not run)

Default model fitting implementation.

Description

Creates a default recipe and formula objects used in model training stage.

Super class

D2MCS::GenericModelFit -> DefaultModelFit

Methods

Public methods

• DefaultModelFit$new()

• DefaultModelFit$createFormula()

• DefaultModelFit$createRecipe()

• DefaultModelFit$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

DefaultModelFit$new()

Method createFormula()

The function is responsible of creating a formula for M.L. model.

Usage

DefaultModelFit$createFormula(instances, class.name, simplify = FALSE)

Arguments

Returns

A formula object.

Method createRecipe()

The function is responsible of creating a recipe with five operations over the data: step_zv, step_nzv, step_corr, step_center, step_scale

Usage

DefaultModelFit$createRecipe(instances, class.name)

Arguments

Details

This function is automatically invoked by D2MCS during model training stage.

Returns

An object of class recipe.

Method clone()

The objects of this class are cloneable with this method.

Usage

DefaultModelFit$clone(deep = FALSE)

Arguments

See Also

GenericModelFit, train

Clustering strategy based on dependency between features.

Description

Features are distributed according to their independence values. This strategy is divided into two steps. The first phase focuses on forming groups with those features most dependent on each other. This step also identifies those that are independent from all the others in the group. The second step is to try out different numbers of clusters until you find the one you think is best. These clusters are formed by inserting in all the independent characteristics identified previously and trying to distribute the features of the groups formed in the previous step in separate clusters. In this way, it seeks to ensure that the features are as independent as possible from those found in the same cluster.

Details

The strategy is suitable only for binary and real features. Other features are automatically grouped into a specific cluster named as 'unclustered'. This class requires the StrategyConfiguration type object implements the following methods:

- getBinaryCutoff(): The function is used to define the interval to consider the dependency between binary features.

- getRealCutoff(): The function allows defining the cutoff to consider the dependency between real features.

- tiebreak(feature, clus.candidates, fea.dep.dist.clus, corpus, heuristic, class, class.name): The function solves the ties between two (or more) features.

- qualityOfCluster(clusters, metrics): The function determines the quality of a cluster

- isImprovingClustering(clusters.deltha): The function indicates if clustering is getting better as the number of them increases.

An example of implementation with the description of each parameter is the DependencyBasedStrategyConfiguration class.

Super class

D2MCS::GenericClusteringStrategy -> DependencyBasedStrategy

Methods

Public methods

• DependencyBasedStrategy$new()

• DependencyBasedStrategy$execute()

• DependencyBasedStrategy$getDistribution()

• DependencyBasedStrategy$createTrain()

• DependencyBasedStrategy$plot()

• DependencyBasedStrategy$saveCSV()

• DependencyBasedStrategy$clone()

Method new()

Method for initializing the object parameters during runtime.

Usage

DependencyBasedStrategy$new(
  subset,
  heuristic,
  configuration = DependencyBasedStrategyConfiguration$new()
)

Arguments

Method execute()

Function responsible of performing the dependency-based feature clustering strategy over the defined Subset.

Usage

DependencyBasedStrategy$execute(verbose = TRUE)

Arguments

Method getDistribution()

Function used to obtain a specific cluster distribution.

Usage

DependencyBasedStrategy$getDistribution(
  num.clusters = NULL,
  num.groups = NULL,
  include.unclustered = FALSE
)

Arguments

Returns

A list with the features comprising an specific clustering distribution.

Method createTrain()

The function is used to create a Trainset object from a specific clustering distribution.

Usage

DependencyBasedStrategy$createTrain(
  subset,
  num.clusters = NULL,
  num.groups = NULL,
  include.unclustered = FALSE
)

Arguments

Details

If num.clusters and num.groups are not defined, best clustering distribution is used to create the train set.

Method plot()

The function is responsible for creating a plot to visualize the clustering distribution.

Usage

DependencyBasedStrategy$plot(dir.path = NULL, file.name = NULL)

Arguments

Method saveCSV()

The function is used to save the clustering distribution to a CSV file.

Usage

DependencyBasedStrategy$saveCSV(
  dir.path = NULL,
  name = NULL,
  num.clusters = NULL
)

Arguments

Method clone()

The objects of this class are cloneable with this method.

Usage

DependencyBasedStrategy$clone(deep = FALSE)

Arguments

See Also

GenericClusteringStrategy, StrategyConfiguration, DependencyBasedStrategyConfiguration

Custom Strategy Configuration handler for the DependencyBasedStrategy strategy.

Description

Define the default configuration parameters for the DependencyBasedStrategy strategy.

Super class

D2MCS::StrategyConfiguration -> DependencyBasedStrategyConfiguration

Methods

Public methods

• DependencyBasedStrategyConfiguration$new()

• DependencyBasedStrategyConfiguration$minNumClusters()

• DependencyBasedStrategyConfiguration$maxNumClusters()

• DependencyBasedStrategyConfiguration$getBinaryCutoff()

• DependencyBasedStrategyConfiguration$getRealCutoff()

• DependencyBasedStrategyConfiguration$setBinaryCutoff()

• DependencyBasedStrategyConfiguration$setRealCutoff()

• DependencyBasedStrategyConfiguration$tiebreak()

• DependencyBasedStrategyConfiguration$qualityOfCluster()

• DependencyBasedStrategyConfiguration$isImprovingClustering()

• DependencyBasedStrategyConfiguration$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

DependencyBasedStrategyConfiguration$new(
  binaryCutoff = 0.6,
  realCutoff = 0.6,
  tiebreakMethod = "lfdc",
  metric = "dep.tar"
)

Arguments

Method minNumClusters()

Function used to return the minimum number of clusters distributions used. By default the minimum is set in 2.

Usage

DependencyBasedStrategyConfiguration$minNumClusters(...)

Arguments

Returns

A numeric vector of length 1.

Method maxNumClusters()

The function is responsible of returning the maximum number of cluster distributions used. By default the maximum number is set in 50.

Usage

DependencyBasedStrategyConfiguration$maxNumClusters(...)

Arguments

Returns

A numeric vector of length 1.

Method getBinaryCutoff()

Gets the cutoff to consider the dependency between binary features.

Usage

DependencyBasedStrategyConfiguration$getBinaryCutoff()

Returns

The numeric value of binary cutoff.

Method getRealCutoff()

Gets the cutoff to consider the dependency between real features.

Usage

DependencyBasedStrategyConfiguration$getRealCutoff()

Returns

The numeric value of real cutoff.

Method setBinaryCutoff()

Sets the cutoff to consider the dependency between binary features.

Usage

DependencyBasedStrategyConfiguration$setBinaryCutoff(cutoff)

Arguments

Method setRealCutoff()

Sets the cutoff to consider the dependency between real features.

Usage

DependencyBasedStrategyConfiguration$setRealCutoff(cutoff)

Arguments

Method tiebreak()

The function solves the ties between two (or more) features.

Usage

DependencyBasedStrategyConfiguration$tiebreak(
  feature,
  clus.candidates,
  fea.dep.dist.clus,
  corpus,
  heuristic,
  class,
  class.name
)

Arguments

Method qualityOfCluster()

The function determines the quality of a cluster.

Usage

DependencyBasedStrategyConfiguration$qualityOfCluster(clusters, metrics)

Arguments

Returns

A numeric vector of length 1.

Method isImprovingClustering()

The function indicates if clustering is getting better as the number of them increases.

Usage

DependencyBasedStrategyConfiguration$isImprovingClustering(clusters.deltha)

Arguments

Returns

A numeric vector of length 1.

Method clone()

The objects of this class are cloneable with this method.

Usage

DependencyBasedStrategyConfiguration$clone(deep = FALSE)

Arguments

See Also

StrategyConfiguration, DependencyBasedStrategy

Handles training of M.L. models

Description

Allows to manage the executed M.L. models.

Methods

Public methods

• ExecutedModels$new()

• ExecutedModels$getNames()

• ExecutedModels$getBest()

• ExecutedModels$add()

• ExecutedModels$exist()

• ExecutedModels$size()

• ExecutedModels$save()

• ExecutedModels$delete()

• ExecutedModels$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

ExecutedModels$new(dir.path)

Arguments

Method getNames()

The function is used to obtain the name of the ML model achieved the best performance during training stage.

Usage

ExecutedModels$getNames()

Returns

A character vector of length 1 of NULL if no ML model have been trained.

Method getBest()

The function is responsible of returning the model achieving the best performance value during training stage.

Usage

ExecutedModels$getBest()

Returns

A Model object.

Method add()

The function inserts a new model to the list of executed models.

Usage

ExecutedModels$add(model, keep.best = TRUE)

Arguments

Method exist()

The function is used to discern if a specific model has been executed previously.

Usage

ExecutedModels$exist(model.name)

Arguments

Returns

A logical value. TRUE if the model exists and FALSE otherwise.

Method size()

The function is used to compute the number of executed ML models.

Usage

ExecutedModels$size()

Returns

A numeric vector or size 1.

Method save()

The function is responsible of saving the information of all executed models into a hidden file.

Usage

ExecutedModels$save()

Method delete()

The function removes an specific model.

Usage

ExecutedModels$delete(model.name)

Arguments

Method clone()

The objects of this class are cloneable with this method.

Usage

ExecutedModels$clone(deep = FALSE)

Arguments

See Also

Model

Iterator over a file.

Description

Creates a FIterator object to iterate over high dimensional files.

Details

Use HDDataset class to ensure the creation of a valid FIterator object.

Methods

Public methods

• FIterator$new()

• FIterator$getNext()

• FIterator$isLast()

• FIterator$finalize()

• FIterator$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

FIterator$new(config.params, chunk.size, verbose)

Arguments

Method getNext()

Gets the next chunk of data. Each iteration returns the same instances (data.frame rows) as chunk.size. However, if remaining data if less than chunk size, all the remaining data is returned. Conversely, NULL when there is no more pending data. By default chunk.size is defined as 10000.

Usage

FIterator$getNext()

Returns

A data.frame of NULL if all the data have been previously returned.

Method isLast()

Checks if the FIterator object reached the end of the data.frame

Usage

FIterator$isLast()

Returns

A logical value indicating if the end of data.frame has been reached.

Method finalize()

Destroys the FIterator object.

Usage

FIterator$finalize()

Method clone()

The objects of this class are cloneable with this method.

Usage

FIterator$clone(deep = FALSE)

Arguments

See Also

Dataset

Computes the False Negative errors.

Description

Computes the ratio of number of Type II errors achieved by the final M.L. model.

Super class

D2MCS::MeasureFunction -> FN

Methods

Public methods

• FN$new()

• FN$compute()

• FN$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

FN$new(performance.output = NULL)

Arguments

Method compute()

The function computes the FN achieved by the M.L. model.

Usage

FN$compute(performance.output = NULL)

Arguments

Details

This function is automatically invoked by the ClassificationOutput framework.

Returns

A numeric vector of size 1 or NULL if an error occurred.

Method clone()

The objects of this class are cloneable with this method.

Usage

FN$clone(deep = FALSE)

Arguments

See Also

MeasureFunction, ClassificationOutput, ConfMatrix

Computes the False Positive value.

Description

This is the number of individuals with a negative condition for which the test result is positive. The value entered here must be non-negative.

Super class

D2MCS::MeasureFunction -> FP

Methods

Public methods

• FP$new()

• FP$compute()

• FP$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

FP$new(performance.output = NULL)

Arguments

Method compute()

The function computes the FP achieved by the M.L. model.

Usage

FP$compute(performance.output = NULL)

Arguments

Details

This function is automatically invoked by the ClassificationOutput object.

Returns

A numeric vector of size 1 or NULL if an error occurred.

Method clone()

The objects of this class are cloneable with this method.

Usage

FP$clone(deep = FALSE)

Arguments

See Also

MeasureFunction, ClassificationOutput, ConfMatrix

Stores the prediction for a specific voting scheme.

Description

The class is used to store the computed probability after executing an specific voting scheme.

Methods

Public methods

• FinalPred$new()

• FinalPred$set()

• FinalPred$getProb()

• FinalPred$getRaw()

• FinalPred$getClassValues()

• FinalPred$getPositiveClass()

• FinalPred$getNegativeClass()

• FinalPred$clone()

Method new()

Method for initializing the object variables during runtime.

Usage

FinalPred$new()

Method set()

Sets the computed probabilities after executing an specific voting scheme.

Usage

FinalPred$set(prob, raw, class.values, positive.class)

Arguments

Method getProb()

Gets the probabilities of the prediction for a specific voting scheme.

Usage

FinalPred$getProb()

Returns

The vector value of probabilities of the prediction for a specific voting scheme.

Method getRaw()

Gets the raw results of the prediction for a specific voting scheme.

Usage

FinalPred$getRaw()

Returns

The vector value of raw results of the prediction for a specific voting scheme.

Method getClassValues()

Gets the class values (positive class + negative class).

Usage

FinalPred$getClassValues()

Returns

The vector value of class values.

Method getPositiveClass()

Gets the positive class.

Usage

FinalPred$getPositiveClass()

Returns

The character value of positive class.

Method getNegativeClass()

Gets the negative class.

Usage

FinalPred$getNegativeClass()

Returns

The character value of negative class.

Method clone()

The objects of this class are cloneable with this method.

Usage

FinalPred$clone(deep = FALSE)

Arguments

See Also

Prediction, SimpleVoting, SingleVoting, CombinedVoting, VotingStrategy

Feature-clustering based on Fisher's Exact Test.

Description

Performs feature-clustering based on Fisher's exact test for testing the null of independence of rows and columns in a contingency table with fixed marginals.

Super class

D2MCS::GenericHeuristic -> FisherTestHeuristic

Methods

Public methods

• FisherTestHeuristic$new()

• FisherTestHeuristic$heuristic()

• FisherTestHeuristic$clone()

Method new()

Empty function used to initialize the object arguments in runtime.

Usage

FisherTestHeuristic$new()

Method heuristic()

Performs the Fisher's exact test for testing the null of independence between two columns (col1 and col2).

Usage

FisherTestHeuristic$heuristic(col1, col2, column.names = NULL)

Arguments

Returns

A numeric vector of length 1 or NA if an error occurs.

Method clone()

The objects of this class are cloneable with this method.

Usage

FisherTestHeuristic$clone(deep = FALSE)

Arguments

See Also

Dataset, fisher.test

Feature-clustering based on GainRatio methodology.

Description

Performs the feature-clustering using entropy-based filters.

Super class

D2MCS::GenericHeuristic -> GainRatioHeuristic

Methods

Public methods

• GainRatioHeuristic$new()

• GainRatioHeuristic$heuristic()

• GainRatioHeuristic$clone()

Method new()

Empty function used to initialize the object arguments in runtime.

Usage

GainRatioHeuristic$new()

Method heuristic()

The algorithms find weights of discrete attributes basing on their correlation with continuous class attribute.

Usage

GainRatioHeuristic$heuristic(col1, col2, column.names = NULL)

Arguments

Returns

A numeric vector of length 1 or NA if an error occurs.

Method clone()

The objects of this class are cloneable with this method.

Usage

GainRatioHeuristic$clone(deep = FALSE)

Arguments

See Also

Dataset, gain.ratio

Abstract Feature Clustering Strategy class.

Description

Abstract class used as a template to ensure the proper definition of new customized clustering strategies.

Details

The GenericClusteringStrategy is an archetype class so it cannot be instantiated.

Methods

Public methods

• GenericClusteringStrategy$new()

• GenericClusteringStrategy$getDescription()

• GenericClusteringStrategy$getHeuristic()

• GenericClusteringStrategy$getConfiguration()

• GenericClusteringStrategy$getBestClusterDistribution()

• GenericClusteringStrategy$getUnclustered()

• GenericClusteringStrategy$execute()

• GenericClusteringStrategy$getDistribution()

• GenericClusteringStrategy$createTrain()

• GenericClusteringStrategy$plot()

• GenericClusteringStrategy$saveCSV()

• GenericClusteringStrategy$clone()

Method new()

A function responsible for creating a GenericClusteringStrategy object.

Usage

GenericClusteringStrategy$new(subset, heuristic, description, configuration)

Arguments

Method getDescription()

The function is used to obtain the description of the strategy.

Usage

GenericClusteringStrategy$getDescription()

Returns

A character vector of NULL if not defined.

Method getHeuristic()

The function returns the heuristic applied for the clustering strategy.

Usage

GenericClusteringStrategy$getHeuristic()

Returns

An object inherited from GenericClusteringStrategy class.

Method getConfiguration()

The function returns the configuration parameters used to perform the clustering strategy.

Usage

GenericClusteringStrategy$getConfiguration()

Returns

An object inherited from StrategyConfiguration class.

Method getBestClusterDistribution()

The function obtains the best clustering distribution.

Usage

GenericClusteringStrategy$getBestClusterDistribution()

Returns

A list of clusters. Each list element represents a feature group.

Method getUnclustered()

The function is used to return the features that cannot be clustered due to incompatibilities with the used heuristic.

Usage

GenericClusteringStrategy$getUnclustered()

Returns

A character vector containing the unclassified features.

Method execute()

Abstract function responsible of performing the clustering strategy over the defined Subset.

Usage

GenericClusteringStrategy$execute(verbose, ...)

Arguments

Method getDistribution()

Abstract function used to obtain the set of features following an specific clustering distribution.

Usage

GenericClusteringStrategy$getDistribution(
  num.clusters = NULL,
  num.groups = NULL,
  include.unclustered = FALSE
)

Arguments

Returns

A list with the features comprising an specific clustering distribution.

Method createTrain()

Abstract function in charge of creating a Trainset object for training purposes.

Usage

GenericClusteringStrategy$createTrain(
  subset,
  num.cluster = NULL,
  num.groups = NULL,
  include.unclustered = FALSE
)

Arguments

Method plot()

Abstract function responsible of creating a plot to visualize the clustering distribution.

Usage

GenericClusteringStrategy$plot(dir.path = NULL, file.name = NULL, ...)

Arguments

Method saveCSV()

Abstract function to save the clustering distribution to a CSV file.

Usage

GenericClusteringStrategy$saveCSV(dir.path, name, num.clusters = NULL)

Arguments

Method clone()

The objects of this class are cloneable with this method.

Usage

GenericClusteringStrategy$clone(deep = FALSE)

Arguments

See Also

Subset, GenericHeuristic

Abstract Feature Clustering heuristic object.

Description

Abstract class used as a template to define new customized clustering heuristics.

Details

The GenericHeuristic is an archetype class so it cannot be instantiated.

Methods

Public methods

• GenericHeuristic$new()

• GenericHeuristic$heuristic()

• GenericHeuristic$clone()

Method new()

Empty function used to initialize the object arguments in runtime.

Usage

GenericHeuristic$new()

Method heuristic()

Function used to implement the clustering heuristic.

Usage

GenericHeuristic$heuristic(col1, col2, column.names = NULL, ...)

Arguments

Returns

A numeric vector of length 1.

Method clone()

The objects of this class are cloneable with this method.

Usage

GenericHeuristic$clone(deep = FALSE)

Arguments

See Also

Dataset

Abstract class for defining model fitting method.

Description

Template to create a recipe or formula objects used in model training stage.

Methods

Public methods

• GenericModelFit$new()

• GenericModelFit$createFormula()

• GenericModelFit$createRecipe()

• GenericModelFit$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

GenericModelFit$new()

Method createFormula()

The function is responsible of creating a formula for M.L. model.

Usage

GenericModelFit$createFormula(instances, class.name, simplify = TRUE)

Arguments

Returns

A formula object.

Method createRecipe()

The function is responsible of creating a recipe for M.L. model.

Usage

GenericModelFit$createRecipe(instances, class.name)

Arguments

Returns

A object of class recipe.

Method clone()

The objects of this class are cloneable with this method.

Usage

GenericModelFit$clone(deep = FALSE)

Arguments

See Also

DefaultModelFit, train

Pseudo-abstract class for creating feature clustering plots.

Description

The GenericPlot implements a basic plot.

Methods

Public methods

• GenericPlot$new()

• GenericPlot$plot()

• GenericPlot$clone()

Method new()

Empty function used to initialize the object arguments in runtime.

Usage

GenericPlot$new()

Method plot()

Implements a generic plot to visualize basic feature-clustering data.

Usage

GenericPlot$plot(summary)

Arguments

Method clone()

The objects of this class are cloneable with this method.

Usage

GenericPlot$clone(deep = FALSE)

Arguments

See Also

BinaryPlot

High Dimensional Dataset handler.

Description

Creates a high dimensional dataset object. Only the required instances are loaded in memory to avoid unnecessary of resources and memory.

Methods

Public methods

• HDDataset$new()

• HDDataset$getColumnNames()

• HDDataset$getNcol()

• HDDataset$createSubset()

Method new()

Method for initializing the object arguments during runtime.

Usage

HDDataset$new(
  filepath,
  header = TRUE,
  sep = ",",
  skip = 0,
  normalize.names = FALSE,
  ignore.columns = NULL
)

Arguments

Method getColumnNames()

Gets the name of the columns comprising the dataset

Usage

HDDataset$getColumnNames()

Returns

A character vector with the name of each column.

Method getNcol()

Obtains the number of columns present in the dataset.

Usage

HDDataset$getNcol()

Returns

An integer of length 1 or NULL

Method createSubset()

Creates a blinded HDSubset for classification purposes.

Usage

HDDataset$createSubset(column.id = FALSE, chunk.size = 1e+05)

Arguments

Returns

A HDSubset object.

See Also

Dataset, HDSubset, DatasetLoader

High Dimensional Subset handler.

Description

Creates a high dimensional subset from a HDDataset object. Only the required instances are loaded in memory to avoid unnecessary use of resources and memory.

Details

Use HDDataset to ensure the creation of a valid HDSubset object.

Methods

Public methods

• HDSubset$new()

• HDSubset$getColumnNames()

• HDSubset$getNcol()

• HDSubset$getID()

• HDSubset$getIterator()

• HDSubset$isBlinded()

• HDSubset$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

HDSubset$new(
  file.path,
  feature.names,
  feature.id,
  start.at = 0,
  sep = ",",
  chunk.size
)

Arguments

Method getColumnNames()

Gets the name of the columns comprising the subset.

Usage

HDSubset$getColumnNames()

Returns

A character vector containing the name of each column.

Method getNcol()

Obtains the number of columns present in the dataset.

Usage

HDSubset$getNcol()

Returns

A numeric value or 0 if is empty.

Method getID()

Obtains the column identifier.

Usage

HDSubset$getID()

Returns

A character vector of size 1.

Method getIterator()

Creates the FIterator object.

Usage

HDSubset$getIterator(chunk.size = private$chunk.size, verbose = FALSE)

Arguments

Returns

A FIterator object to transverse through HDSubset instances

Method isBlinded()

Checks if the subset contains a target class.

Usage

HDSubset$isBlinded()

Returns

A logical to specify if the subset contains a target class or not.

Method clone()

The objects of this class are cloneable with this method.

Usage

HDSubset$clone(deep = FALSE)

Arguments

See Also

HDDataset, DatasetLoader

Feature-clustering based on InformationGain methodology.

Description

Performs the feature-clustering using entropy-based filters.

Super class

D2MCS::GenericHeuristic -> InformationGainHeuristic

Methods

Public methods

• InformationGainHeuristic$new()

• InformationGainHeuristic$heuristic()

• InformationGainHeuristic$clone()

Method new()

Empty function used to initialize the object arguments in runtime.

Usage

InformationGainHeuristic$new()

Method heuristic()

The algorithm find weights of discrete attributes basing on their correlation with continuous class attribute. Particularly Information Gain uses H(Class) + H(Attribute) - H(Class, Attribute)

Usage

InformationGainHeuristic$heuristic(col1, col2, column.names = NULL)

Arguments

Returns

A numeric vector of length 1 or NA if an error occurs.

Method clone()

The objects of this class are cloneable with this method.

Usage

InformationGainHeuristic$clone(deep = FALSE)

Arguments

See Also

Dataset, information.gain

Computes the Kappa Cohen value.

Description

Cohen's Kappa measures the agreement between two raters who each classify N items into C mutually exclusive categories.

Details

\kappa \hspace{0.1cm} is \hspace{0.1cm} equivalent \hspace{0.1cm} to \hspace{0.1cm} (p_o - p_e) / (1 - p_e) = 1 - (1 - p_0) / (1 - p_e)

Super class

D2MCS::MeasureFunction -> Kappa

Methods

Public methods

• Kappa$new()

• Kappa$compute()

• Kappa$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

Kappa$new(performance.output = NULL)

Arguments

Method compute()

The function computes the Kappa achieved by the M.L. model.

Usage

Kappa$compute(performance.output = NULL)

Arguments

Details

This function is automatically invoked by the ClassificationOutput object.

Returns

A numeric vector of size 1 or NULL if an error occurred.

Method clone()

The objects of this class are cloneable with this method.

Usage

Kappa$clone(deep = FALSE)

Arguments

See Also

MeasureFunction, ClassificationOutput, ConfMatrix

Feature-clustering based on Kendall Correlation Test.

Description

Performs the feature-clustering using Kendall correlation tests.

Details

The method estimate the association between paired samples and compute a test of the value being zero. They use different measures of association, all in the range [-1, 1] with 0 indicating no association. Method valid only for bi-class problems.

Super class

D2MCS::GenericHeuristic -> KendallHeuristic

Methods

Public methods

• KendallHeuristic$new()

• KendallHeuristic$heuristic()

• KendallHeuristic$clone()

Method new()

Empty function used to initialize the object arguments in runtime.

Usage

KendallHeuristic$new()

Method heuristic()

Test for association between paired samples using Kendall's tau value.

Usage

KendallHeuristic$heuristic(col1, col2, column.names = NULL)

Arguments

Returns

a numeric vector of length 1 or NA if an error occurs.

Method clone()

The objects of this class are cloneable with this method.

Usage

KendallHeuristic$clone(deep = FALSE)

Arguments

See Also

Dataset, cor.test

Computes the Matthews correlation coefficient.

Description

The Matthews correlation coefficient is used in machine learning as a measure of the quality of binary (two-class) classifications. It takes into account true and false positives and negatives and is generally regarded as a balanced measure which can be used even if the classes are of very different sizes. The MCC is in essence a correlation coefficient between the observed and predicted binary classifications; it returns a value between -1 and +1.

Details

MCC = (TP × (TN - FP) × FN)/(\sqrt{(TP + FP) × (TP + FN) × (TN + FP) × (TN + FN)})

Super class

D2MCS::MeasureFunction -> MCC

Methods

Public methods

• MCC$new()

• MCC$compute()

• MCC$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

MCC$new(performance.output = NULL)

Arguments

Method compute()

The function computes the MCC achieved by the M.L. model.

Usage

MCC$compute(performance.output = NULL)

Arguments

Details

This function is automatically invoke by the ClassificationOutput object.

Returns

A numeric vector of size 1 or NULL if an error occurred.

Method clone()

The objects of this class are cloneable with this method.

Usage

MCC$clone(deep = FALSE)

Arguments

See Also

MeasureFunction, ClassificationOutput, ConfMatrix

Feature-clustering based on Matthews Correlation Coefficient score.

Description

Performs the feature-clustering using MCC score. Valid for both bi-class and multi-class problems

Super class

D2MCS::GenericHeuristic -> MCCHeuristic

Methods

Public methods

• MCCHeuristic$new()

• MCCHeuristic$heuristic()

• MCCHeuristic$clone()

Method new()

Empty function used to initialize the object arguments in runtime.

Usage

MCCHeuristic$new()

Method heuristic()

Calculates the Matthews correlation Coefficient (MCC) score.

Usage

MCCHeuristic$heuristic(col1, col2, column.names = NULL)

Arguments

Returns

A numeric vector of length 1 or NA if an error occurs.

Method clone()

The objects of this class are cloneable with this method.

Usage

MCCHeuristic$clone(deep = FALSE)

Arguments

See Also

Dataset, mccr

Archetype to define customized measures.

Description

Abstract class used as a template to define new M.L. performance measures.

Details

The GenericHeuristic is an full-abstract class so it cannot be instantiated. To ensure the proper operation, compute method is automatically invoke by D2MCS framework when needed.

Methods

Public methods

• MeasureFunction$new()

• MeasureFunction$compute()

• MeasureFunction$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

MeasureFunction$new(performance = NULL)

Arguments

Method compute()

The function implements the metric used to measure the performance achieved by the M.L. model.

Usage

MeasureFunction$compute(performance.output = NULL)

Arguments

Details

This function is automatically invoke by the D2MCS framework.

Returns

A numeric vector of size 1 or NULL if an error occurred.

Method clone()

The objects of this class are cloneable with this method.

Usage

MeasureFunction$clone(deep = FALSE)

Arguments

See Also

MeasureFunction

Abstract class to compute the probability prediction based on combination between metrics.

Description

Abstract class used as a template to define new customized strategies to combine the probability predictions made by different metrics.

Methods

Public methods

• Methodology$new()

• Methodology$getRequiredMetrics()

• Methodology$compute()

• Methodology$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

Methodology$new(required.metrics)

Arguments

Method getRequiredMetrics()

The function returns the required metrics that will participate in the methodology to compute a metric based on all of them.

Usage

Methodology$getRequiredMetrics()

Returns

A character vector of length greater than 2 with the name of the required metrics.

Method compute()

Function to compute the probability of the final prediction based on different metrics.

Usage

Methodology$compute(raw.pred, prob.pred, positive.class, negative.class)

Arguments

Returns

A numeric value indicating the probability of the instance is predicted as positive class.

Method clone()

The objects of this class are cloneable with this method.

Usage

Methodology$clone(deep = FALSE)

Arguments

See Also

ProbBasedMethodology

Combined metric strategy to minimize FN errors.

Description

Calculates if the positive class is the predicted one in any of the metrics, otherwise, the instance is not considered to have the positive class associated.

Super class

D2MCS::CombinedMetrics -> MinimizeFN

Methods

Public methods

• MinimizeFN$new()

• MinimizeFN$getFinalPrediction()

• MinimizeFN$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

MinimizeFN$new(required.metrics = c("MCC", "PPV"))

Arguments

Method getFinalPrediction()

Function to obtain the final prediction based on different metrics.

Usage

MinimizeFN$getFinalPrediction(
  raw.pred,
  prob.pred,
  positive.class,
  negative.class
)

Arguments

Returns

A logical value indicating if the instance is predicted as positive class or not.

Method clone()

The objects of this class are cloneable with this method.

Usage

MinimizeFN$clone(deep = FALSE)

Arguments

See Also

CombinedMetrics

Combined metric strategy to minimize FP errors.

Description

Calculates if the positive class is the predicted one in all metrics, otherwise, the instance is not considered to have the positive class associated.

Super class

D2MCS::CombinedMetrics -> MinimizeFP

Methods

Public methods

• MinimizeFP$new()

• MinimizeFP$getFinalPrediction()

• MinimizeFP$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

MinimizeFP$new(required.metrics = c("MCC", "PPV"))

Arguments

Method getFinalPrediction()

Function to obtain the final prediction based on different metrics.

Usage

MinimizeFP$getFinalPrediction(
  raw.pred,
  prob.pred,
  positive.class,
  negative.class
)

Arguments

Returns

A logical value indicating if the instance is predicted as positive class or not.

Method clone()

The objects of this class are cloneable with this method.

Usage

MinimizeFP$clone(deep = FALSE)

Arguments

See Also

CombinedMetrics

Stores a previously trained M.L. model.

Description

Encapsulates and handles all the information and operations associated with a M.L. model.

Methods

Public methods

• Model$new()

• Model$isTrained()

• Model$getDir()

• Model$getName()

• Model$getFamily()

• Model$getDescription()

• Model$train()

• Model$getTrainedModel()

• Model$getExecutionTime()

• Model$getPerformance()

• Model$getConfiguration()

• Model$save()

• Model$remove()

• Model$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

Model$new(dir.path, model)

Arguments

Method isTrained()

The function is used to determine is the model has been already trained.

Usage

Model$isTrained()

Returns

A logical value. TRUE if the model has been trained and FALSE otherwise.

Method getDir()

The function returns the location path of the specific model.

Usage

Model$getDir()

Returns

A character vector specifying the location of the model.

Method getName()

The function is used to obtain the name of the model.

Usage

Model$getName()

Returns

A character vector with the name of the model.

Method getFamily()

The function gets the family of the model.

Usage

Model$getFamily()

Returns

A character vector representing the family of the ML model.

Method getDescription()

The function allows obtaining the description associated with an specific ML model.

Usage

Model$getDescription()

Returns

A character vector with the model description.

Method train()

The function is responsible of performing model training operation.

Usage

Model$train(train.set, fitting, trFunction, metric, logs)

Arguments

Method getTrainedModel()

The function allows obtaining the trained model.

Usage

Model$getTrainedModel()

Returns

A train class.

Method getExecutionTime()

The function is used to compute the time taken to perform training operation.

Usage

Model$getExecutionTime()

Returns

A numeric vector with length 1.

Method getPerformance()

The function obtains the performance achieved by the model during training stage.

Usage

Model$getPerformance(metric = private$metric)

Arguments

Returns

A numeric value with the performance achieved.

Method getConfiguration()

The function is used to get the configuration parameters achieved by the ML model after the training stage.

Usage

Model$getConfiguration()

Returns

A list object with the configuration parameters.

Method save()

The function is responsible of saving the model to disc into a RDS file.

Usage

Model$save(replace = TRUE)

Arguments

Method remove()

The function is used to delete a model from disc.

Usage

Model$remove()

Method clone()

The objects of this class are cloneable with this method.

Usage

Model$clone(deep = FALSE)

Arguments

See Also

D2MCS, TrainFunction

Feature-clustering based on Mutual Information Computation theory.

Description

Performs the feature-clustering using MCC score. Valid for both bi-class and multi-class problems. Only valid for bi-class problems.

Super class

D2MCS::GenericHeuristic -> MultinformationHeuristic

Methods

Public methods

• MultinformationHeuristic$new()

• MultinformationHeuristic$heuristic()

• MultinformationHeuristic$clone()

Method new()

Empty function used to initialize the object arguments in runtime.

Usage

MultinformationHeuristic$new()

Method heuristic()

Mutinformation takes two random variables as input and computes the mutual information in nats according to the entropy estimator method.

Usage

MultinformationHeuristic$heuristic(col1, col2, column.names = NULL)

Arguments

Returns

Returns the mutual information I(X;Y) in nats.

Method clone()

The objects of this class are cloneable with this method.

Usage

MultinformationHeuristic$clone(deep = FALSE)

Arguments

See Also

Dataset, mutinformation

Computes the Negative Predictive Value.

Description

Negative Predictive Values are the proportions of negative results in statistics and diagnostic tests that are true negative results.

Details

NPV = TN / (TN + FN)

Super class

D2MCS::MeasureFunction -> NPV

Methods

Public methods

• NPV$new()

• NPV$compute()

• NPV$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

NPV$new(performance.output = NULL)

Arguments

Method compute()

The function computes the NPV achieved by the M.L. model.

Usage

NPV$compute(performance.output = NULL)

Arguments

Details

This function is automatically invoke by the ClassificationOutput object.

Returns

A numeric vector of size 1 or NULL if an error occurred.

Method clone()

The objects of this class are cloneable with this method.

Usage

NPV$clone(deep = FALSE)

Arguments

See Also

MeasureFunction, ClassificationOutput, ConfMatrix

Compute performance across resamples.

Description

Computes the performance across resamples when class probabilities cannot be computed.

Super class

D2MCS::SummaryFunction -> NoProbability

Methods

Public methods

• NoProbability$new()

• NoProbability$execute()

• NoProbability$clone()

Method new()

The function defined during runtime the usage of five measures: 'Kappa', 'Accuracy', 'TCR_9', 'MCC' and 'PPV'.

Usage

NoProbability$new()

Method execute()

The function computes the performance across resamples using the previously defined measures.

Usage

NoProbability$execute(data, lev = NULL, model = NULL)

Arguments

Returns

A vector of performance estimates.

Method clone()

The objects of this class are cloneable with this method.

Usage

NoProbability$clone(deep = FALSE)

Arguments

See Also

SummaryFunction

Feature-clustering based on Odds Ratio measure.

Description

Performs the feature-clustering using Odds Ratio methodology. Valid only for bi-class problems.

Super class

D2MCS::GenericHeuristic -> OddsRatioHeuristic

Methods

Public methods

• OddsRatioHeuristic$new()

• OddsRatioHeuristic$heuristic()

• OddsRatioHeuristic$clone()

Method new()

Empty function used to initialize the object arguments in runtime.

Usage

OddsRatioHeuristic$new()

Method heuristic()

Calculates the Odds Ratio method.

Usage

OddsRatioHeuristic$heuristic(col1, col2, column.names = NULL)

Arguments

Returns

A numeric vector of length 1 or NA if an error occurs.

Method clone()

The objects of this class are cloneable with this method.

Usage

OddsRatioHeuristic$clone(deep = FALSE)

Arguments

See Also

Dataset, odds.ratio

Computes the Positive Predictive Value.

Description

Positive Predictive Values are the proportions of positive results in statistics and diagnostic tests that are true positive results.

Details

PPV = TP / (TP + FP)

Super class

D2MCS::MeasureFunction -> PPV

Methods

Public methods

• PPV$new()

• PPV$compute()

• PPV$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

PPV$new(performance.output = NULL)

Arguments

Method compute()

The function computes the PPV achieved by the M.L. model.

Usage

PPV$compute(performance.output = NULL)

Arguments

Details

This function is automatically invoke by the ClassificationOutput object.

Returns

A numeric vector of size 1 or NULL if an error occurred.

Method clone()

The objects of this class are cloneable with this method.

Usage

PPV$clone(deep = FALSE)

Arguments

See Also

MeasureFunction, ClassificationOutput, ConfMatrix

Feature-clustering based on Pearson Correlation Test.

Description

Performs the feature-clustering using Pearson correlation tests. Valid for both, bi-class and multi-class problems.

Details

The test statistic is based on Pearson's product moment correlation coefficient cor(x, y) and follows a t distribution with length(x)-2 degrees of freedom if the samples follow independent normal distributions. If there are at least 4 complete pairs of observation, an asymptotic confidence interval is given based on Fisher's Z transform.

Super class

D2MCS::GenericHeuristic -> PearsonHeuristic

Methods

Public methods

• PearsonHeuristic$new()

• PearsonHeuristic$heuristic()

• PearsonHeuristic$clone()

Method new()

Creates a PearsonHeuristic object.

Usage

PearsonHeuristic$new()

Method heuristic()

Test for association between paired samples using Pearson test.

Usage

PearsonHeuristic$heuristic(col1, col2, column.names = NULL)

Arguments

Returns

A numeric vector of length 1 or NA if an error occurs.

Method clone()

The objects of this class are cloneable with this method.

Usage

PearsonHeuristic$clone(deep = FALSE)

Arguments

See Also

Dataset, cor

Computes the Precision Value.

Description

Precision is the fraction of relevant instances among the retrieved instances

Details

precision = TP / (TP + FP)

Super class

D2MCS::MeasureFunction -> Precision

Methods

Public methods

• Precision$new()

• Precision$compute()

• Precision$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

Precision$new(performance.output = NULL)

Arguments

Method compute()

The function computes the Precision achieved by the M.L. model.

Usage

Precision$compute(performance.output = NULL)

Arguments

Details

This function is automatically invoke by the ClassificationOutput object.

Returns

A numeric vector of size 1 or NULL if an error occurred.

Method clone()

The objects of this class are cloneable with this method.

Usage

Precision$clone(deep = FALSE)

Arguments

See Also

MeasureFunction, ClassificationOutput, ConfMatrix

Manages the prediction computed for a specific model.

Description

Allows to obtain predictions from the data provided using a pre-trained model.

Methods

Public methods

• Prediction$new()

• Prediction$execute()

• Prediction$getPrediction()

• Prediction$getModelName()

• Prediction$getModelPerformance()

• Prediction$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

Prediction$new(model, feature.id = NULL)

Arguments

Method execute()

Calculates predictions of the values passed by parameters using the corresponding model.

Usage

Prediction$execute(pred.values, class.values, positive.class)

Arguments

Method getPrediction()

The function is used to return the prediction values computed.

Usage

Prediction$getPrediction(type = NULL, target = NULL)

Arguments

Returns

A data.frame with the computed prediction.

Method getModelName()

Gets the model name.

Usage

Prediction$getModelName()

Returns

The character value of model value.

Method getModelPerformance()

Gets the performance of the model.

Usage

Prediction$getModelPerformance()

Returns

The numeric value of the model's performance.

Method clone()

The objects of this class are cloneable with this method.

Usage

Prediction$clone(deep = FALSE)

Arguments

See Also

ClusterPredictions

Encapsulates the achieved predictions.

Description

The class used to encapsulates all the computed predictions to facilitate their access and maintenance.

Methods

Public methods

• PredictionOutput$new()

• PredictionOutput$getPredictions()

• PredictionOutput$getType()

• PredictionOutput$getTarget()

• PredictionOutput$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

PredictionOutput$new(predictions, type, target)

Arguments

Method getPredictions()

The function returns the final predictions.

Usage

PredictionOutput$getPredictions()

Returns

A list containing the final predictions or NULL if classification stage was not successfully performed.

Method getType()

The function returns the type of prediction should be returned. If "prob" or "raw" is defined then computed 'probabilistic' or 'class' values are returned.

Usage

PredictionOutput$getType()

Returns

A character value.

Method getTarget()

The function returns the value of the target class.

Usage

PredictionOutput$getTarget()

Returns

A character value.

Method clone()

The objects of this class are cloneable with this method.

Usage

PredictionOutput$clone(deep = FALSE)

Arguments

See Also

D2MCS

Implementation of Probabilistic Average voting.

Description

Computes the final prediction by performing the mean value of the probability achieved by each prediction.

Super class

D2MCS::SimpleVoting -> ProbAverageVoting

Methods

Public methods

• ProbAverageVoting$new()

• ProbAverageVoting$getMajorityClass()

• ProbAverageVoting$getClassTie()

• ProbAverageVoting$execute()

• ProbAverageVoting$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

ProbAverageVoting$new(cutoff = 0.5, class.tie = NULL, majority.class = NULL)

Arguments

Method getMajorityClass()

The function returns the value of the majority class.

Usage

ProbAverageVoting$getMajorityClass()

Returns

A character vector of length 1 with the name of the majority class.

Method getClassTie()

The function gets the class value assigned to solve ties.

Usage

ProbAverageVoting$getClassTie()

Returns

A character vector of length 1.

Method execute()

The function implements the majority voting procedure.

Usage

ProbAverageVoting$execute(predictions, verbose = FALSE)

Arguments

Method clone()

The objects of this class are cloneable with this method.

Usage

ProbAverageVoting$clone(deep = FALSE)

Arguments

See Also

D2MCS, ClassMajorityVoting, ClassWeightedVoting, ProbAverageVoting, ProbAverageWeightedVoting, ProbBasedMethodology

Implementation of Probabilistic Average Weighted voting.

Description

Computes the final prediction by performing the weighted mean of the probability achieved by each cluster prediction. By default, weight values are consistent with the performance value achieved by the best M.L. model on each cluster.

Super class

D2MCS::SimpleVoting -> ProbAverageWeightedVoting

Methods

Public methods

• ProbAverageWeightedVoting$new()

• ProbAverageWeightedVoting$getClassTie()

• ProbAverageWeightedVoting$getWeights()

• ProbAverageWeightedVoting$setWeights()

• ProbAverageWeightedVoting$execute()

• ProbAverageWeightedVoting$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

ProbAverageWeightedVoting$new(cutoff = 0.5, class.tie = NULL, weights = NULL)

Arguments

Method getClassTie()

The function gets the class value assigned to solve ties.

Usage

ProbAverageWeightedVoting$getClassTie()

Returns

A character vector of length 1.

Method getWeights()

The function returns the value of the majority class.

Usage

ProbAverageWeightedVoting$getWeights()

Returns

A character vector of length 1 with the name of the majority class.

Method setWeights()

The function allows changing the value of the weights.

Usage

ProbAverageWeightedVoting$setWeights(weights)

Arguments

Method execute()

The function implements the cluster-weighted probabilistic voting procedure.

Usage

ProbAverageWeightedVoting$execute(predictions, verbose = FALSE)

Arguments

Method clone()

The objects of this class are cloneable with this method.

Usage

ProbAverageWeightedVoting$clone(deep = FALSE)

Arguments

See Also

D2MCS, ClassMajorityVoting, ClassWeightedVoting, ProbAverageVoting, ProbAverageWeightedVoting, ProbBasedMethodology

Methodology to obtain the combination of the probability of different metrics.

Description

Calculates the mean of the probabilities of the different metrics.

Super class

D2MCS::Methodology -> ProbBasedMethodology

Methods

Public methods

• ProbBasedMethodology$new()

• ProbBasedMethodology$compute()

• ProbBasedMethodology$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

ProbBasedMethodology$new(required.metrics = c("MCC", "PPV"))

Arguments

Method compute()

Function to compute the probability of the final prediction based on different metrics.

Usage

ProbBasedMethodology$compute(
  raw.pred,
  prob.pred,
  positive.class,
  negative.class
)

Arguments

Returns

A numeric value indicating the probability of the instance is predicted as positive class.

Method clone()

The objects of this class are cloneable with this method.

Usage

ProbBasedMethodology$clone(deep = FALSE)

Arguments

See Also

Methodology

Computes the Recall Value.

Description

Recall (also known as sensitivity) is the fraction of the total amount of relevant instances that were actually retrieved.

Details

recall = TP / (TP + FN)

Super class

D2MCS::MeasureFunction -> Recall

Methods

Public methods

• Recall$new()

• Recall$compute()

• Recall$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

Recall$new(performance.output = NULL)

Arguments

Method compute()

The function computes the Recall achieved by the M.L. model.

Usage

Recall$compute(performance.output = NULL)

Arguments

Details

This function is automatically invoke by the ClassificationOutput object.

Returns

A numeric vector of size 1 or NULL if an error occurred.

Method clone()

The objects of this class are cloneable with this method.

Usage

Recall$clone(deep = FALSE)

Arguments

See Also

MeasureFunction, ClassificationOutput, ConfMatrix

Computes the Sensitivity Value.

Description

Sensitivity is a measure of the proportion of actual positive cases that got predicted as positive (or true positive).

Details

Sensitivity = TP / (TP + FN)

Super class

D2MCS::MeasureFunction -> Sensitivity

Methods

Public methods

• Sensitivity$new()

• Sensitivity$compute()

• Sensitivity$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

Sensitivity$new(performance.output = NULL)

Arguments

Method compute()

The function computes the Sensitivity achieved by the M.L. model.

Usage

Sensitivity$compute(performance.output = NULL)

Arguments

Details

This function is automatically invoke by the ClassificationOutput object.

Returns

A numeric vector of size 1 or NULL if an error occurred.

Method clone()

The objects of this class are cloneable with this method.

Usage

Sensitivity$clone(deep = FALSE)

Arguments

See Also

MeasureFunction, ClassificationOutput, ConfMatrix

Simple feature clustering strategy.

Description

Features are sorted by descendant according to the relevance value obtained after applying an specific heuristic. Next, features are distributed into N clusters following a card-dealing methodology. Finally best distribution is assigned to the distribution having highest homogeneity.

Details

The strategy is suitable for all features that are valid for the indicated heuristics. Invalid features are automatically grouped into a specific cluster named as 'unclustered'.

Super class

D2MCS::GenericClusteringStrategy -> SimpleStrategy

Methods

Public methods

• SimpleStrategy$new()

• SimpleStrategy$execute()

• SimpleStrategy$getBestClusterDistribution()

• SimpleStrategy$getUnclustered()

• SimpleStrategy$getDistribution()

• SimpleStrategy$createTrain()

• SimpleStrategy$plot()

• SimpleStrategy$saveCSV()

• SimpleStrategy$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

SimpleStrategy$new(
  subset,
  heuristic,
  configuration = StrategyConfiguration$new()
)

Arguments

Method execute()

Function responsible of performing the clustering strategy over the defined Subset.

Usage

SimpleStrategy$execute(verbose = FALSE)

Arguments

Method getBestClusterDistribution()

The function obtains the best clustering distribution.

Usage

SimpleStrategy$getBestClusterDistribution()

Returns

A list of clusters. Each list element represents a feature group.

Method getUnclustered()

The function is used to return the features that cannot be clustered due to incompatibilities with the used heuristic.

Usage

SimpleStrategy$getUnclustered()

Returns

A character vector containing the unclassified features.

Method getDistribution()

Function used to obtain a specific cluster distribution.

Usage

SimpleStrategy$getDistribution(
  num.clusters = NULL,
  num.groups = NULL,
  include.unclustered = FALSE
)

Arguments

Returns

A list with the features comprising an specific clustering distribution.

Method createTrain()

The function is used to create a Trainset object from a specific clustering distribution.

Usage

SimpleStrategy$createTrain(
  subset,
  num.clusters = NULL,
  num.groups = NULL,
  include.unclustered = FALSE
)

Arguments

Details

If num.clusters and num.groups are not defined, best clustering distribution is used to create the train set.

Returns

A Trainset object.

Method plot()

The function is responsible for creating a plot to visualize the clustering distribution.

Usage

SimpleStrategy$plot(dir.path = NULL, file.name = NULL)

Arguments

Method saveCSV()

The function is used to save the clustering distribution to a CSV file.

Usage

SimpleStrategy$saveCSV(dir.path, name = NULL, num.clusters = NULL)

Arguments

Method clone()

The objects of this class are cloneable with this method.

Usage

SimpleStrategy$clone(deep = FALSE)

Arguments

See Also

GenericClusteringStrategy, StrategyConfiguration

Abtract class to define simple voting schemes.

Description

Abstract class used as a template to define new customized simple voting schemes.

Methods

Public methods

• SimpleVoting$new()

• SimpleVoting$getCutoff()

• SimpleVoting$getFinalPred()

• SimpleVoting$execute()

• SimpleVoting$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

SimpleVoting$new(cutoff = NULL)

Arguments

Method getCutoff()

The function obtains the minimum probabilistic value used to perform a positive classification.

Usage

SimpleVoting$getCutoff()

Returns

A numeric value.

Method getFinalPred()

The function is used to return the prediction values computed by a voting strategy.

Usage

SimpleVoting$getFinalPred(type = NULL, target = NULL, filter = NULL)

Arguments

Returns

A FinalPred object.

Method execute()

Abstract function used to implement the operation of the voting scheme.

Usage

SimpleVoting$execute(predictions, verbose = FALSE)

Arguments

Method clone()

The objects of this class are cloneable with this method.

Usage

SimpleVoting$clone(deep = FALSE)

Arguments

See Also

D2MCS, ClassMajorityVoting, ClassWeightedVoting, ProbAverageVoting, ProbAverageWeightedVoting, ProbBasedMethodology, CombinedVoting

Manages the execution of Simple Votings.

Description

The class is responsible of initializing and executing voting schemes. Additionally, to ensure a proper operation, the class automatically checks the compatibility of defined voting schemes.

Super class

D2MCS::VotingStrategy -> SingleVoting

Methods

Public methods

• SingleVoting$new()

• SingleVoting$execute()

• SingleVoting$clone()

Method new()

The function initializes the object arguments during runtime.

Usage

SingleVoting$new(voting.schemes, metrics)

Arguments

Method execute()

The function is used to execute all the previously defined (and compatible) voting schemes.

Usage

SingleVoting$execute(predictions, verbose = FALSE)

Arguments

Method clone()

The objects of this class are cloneable with this method.

Usage

SingleVoting$clone(deep = FALSE)

Arguments

See Also

D2MCS, SimpleVoting, CombinedVoting

Feature-clustering based on Spearman Correlation Test.

Description

Performs the feature-clustering using Spearman's rho statistic.

Details

Spearman's rho statistic is to estimate a rank-based measure of association. These tests may be used if the data do not necessarily come from a bivariate normal distribution.

Super class

D2MCS::GenericHeuristic -> SpearmanHeuristic

Methods

Public methods

• SpearmanHeuristic$new()

• SpearmanHeuristic$heuristic()

• SpearmanHeuristic$clone()

Method new()

Creates a SpearmanHeuristic object.

Usage

SpearmanHeuristic$new()

Method heuristic()

Test for correlation between paired samples using Spearman rho statistic.

Usage

SpearmanHeuristic$heuristic(col1, col2, column.names = NULL)

Arguments

Returns

A numeric vector of length 1 or NA if an error occurs.

Method clone()

The objects of this class are cloneable with this method.

Usage

SpearmanHeuristic$clone(deep = FALSE)

Arguments

See Also

Dataset, cor.test

Computes the Specificity Value.

Description

Specificity is defined as the proportion of actual negatives, which got predicted as the negative (or true negative). This implies that there will be another proportion of actual negative, which got predicted as positive and could be termed as false positives.

Details

Specificity = True Negative / (True Negative + False Positive)

Super class

D2MCS::MeasureFunction -> Specificity

Methods

Public methods

• Specificity$new()

• Specificity$compute()

• Specificity$clone()

Method new()

Method for initializing the object arguments during runtime.

Usage

Specificity$new(performance.output = NULL)

Arguments

Method compute()

The function computes the Specificity achieved by the M.L. model.

Usage

Specificity$compute(performance.output = NULL)

Arguments

Details

This function is automatically invoke by the ClassificationOutput object.

Returns

A numeric vector of size 1 or NULL if an error occurred.

Method clone()

The objects of this class are cloneable with this method.

Usage

Specificity$clone(deep = FALSE)

Arguments