Coerce to an Edge List

Description

Converts from a Subcascades object to a weighted edge list.

Usage

as.edgedataframe(subcascades)

Arguments

Details

Converts a Subcascades object to a data.frame that can be used to generate a graph.

Value

A data.frame that can be used to generate a graph. The first and second column correspond to all pairwise relations (from - to) of the cascades within the Subcascades object. The 'CASC_ID' column contains the same ID for all edges belonging to the same cascade. The 'SIZE' column gives the size of the cascade to which the repective column bleongs. The method returns NULL if the object subcascades is empty.

See Also

groupwise, as.subcascades

Examples

library(TunePareto)
library(igraph)
data(esl)
data = esl$data
labels = esl$labels
predMap = predictionMap(data, labels, 
                       classifier = tunePareto.svm(), kernel='linear')

# generate a dataframe
subcascades = subcascades(predMap,thresh=0.65,size=4)
edges = as.edgedataframe(subcascades)
g = graph_from_data_frame(edges[,c(1,2)], directed = TRUE)
E(g)$weight = edges[,3]
plot(g,edge.color=edges[,3],edge.arrow.size=0.5,
     edge.curved =seq(-0.5, 1, length = ecount(g)))

Coerce to a Subcascades Object

Description

Converts from a Groupwise object to a Subcascades object.

Usage

as.subcascades(groupwise = NULL)

Arguments

Details

Converts a Groupwise object to a Subcascades object.

Value

A Subcascades object comprising the evaluated cascades and their performances. The Subcascades object is made up of a list of matrices. Each matrix comprises the evaluation results of cascades of a specific length and is sorted row-wise according to the achieved minimal classwise sensitivities of the cascades (decreasing). The rownames show the class order by a character string of type '1>2>3' and the entries the sensitivity for each position of the cascade.

See Also

groupwise, as.edgedataframe

Examples

library(TunePareto)
data(esl)
data = esl$data
labels = esl$labels
foldList = generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap = predictionMap(data, labels, foldList = foldList, 
                       classifier = tunePareto.svm(), kernel='linear')

# generate a Groupwise object
subc = subcascades(predMap,thresh=0.7)
groupwise = groupwise(subc)

#convert it to a Subcascades object
converted.subcascades = as.subcascades(groupwise)

Construction of Binary Classifier Sensitivities

Description

Sensitivities of all pairwise binary classifiers for all classes.

Usage

conf(predictionMap = NULL)

Arguments

Details

The Conf object contains all class sensitivities for each binary trained classifiers. The $fC-part is a matrix with one column and contains the sensitivities for the first class of each pairwise classifier. The rows stand for the pairwise classifiers, whereby 0vs1 means that this classifier was trained for class 0 against class 1, with class 0 being the first class. The number '-1' is used as placeholder. The $sC-part is a matrix and contains the preformance measures for all second classes, which are meant here as all classes except the first class. The rows correspond to the binary classifiers and the columns to the classes.

Value

Object of class Conf. Consists of a list of two numeric matrices fC and sC.

See Also

summary.Conf, print.Conf, plot.Conf

Examples

library(TunePareto)
data(esl)
data = esl$data
labels = esl$labels
foldList = generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap = predictionMap(data, labels, foldList = foldList, 
                       classifier = tunePareto.svm(), kernel='linear')

conf = conf(predMap)

Construction of a Confusion Table

Description

Confusion table and class assignments of one cascade.

Usage

confusionTable(
  predictionMap = NULL,
  cascade = NULL,
  other.classes = NULL,
  sort = TRUE
)

Arguments

Value

An object of type ConfusionTable including the sensitivities, with the label of the predicted classes in the rows and the labels of the original class in the columns.

See Also

summary.ConfusionTable, print.ConfusionTable, plot.ConfusionTable

Examples

library(TunePareto)
data(esl)
data = esl$data
labels = esl$labels
foldList = generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap = predictionMap(data, labels, foldList = foldList, 
                       classifier = tunePareto.svm(), kernel='linear')

# Calculation of the confusion matrix for '0>2>3>4'.
confTable = confusionTable(predMap, cascade = '0>2>3>4')
# Calculation of the confusion matrix for '0>2>3>4' 
# and the assignment of all samples of the other classes.
confTable = confusionTable(predMap, cascade = '0>2>3>4', 
                other.classes='all', sort = TRUE)

Filter out Subcascades

Description

dropSets filters out specific cascade sets.

Usage

dropSets(
  subcascades = NULL,
  sets = NULL,
  direction = "sub",
  ordered = FALSE,
  neighborhood = "direct",
  type = "any"
)

Arguments

Details

This function allows for filtering the Subcascades object.

If direction is set to 'exact' the parameter neighborhood is ignored and the parameter type is used as its default and cannot be changed. There has to be an exact match between the classes of the sets parameter cascades and the cascade of the Subcascades object, so not more but also not less classes are allowed. If the ordered parameter is set to TRUE also the order of the classes within the sets parameter and within the Subcascades object has be be exactly the same.

If direction is set to 'sub' the Subcascades object is filtered for subsets of the given cascades. If the type parameter is set to 'any' each cascade is considered individually otherwise all cascades are used as reference for filtering. This means that either for each cascade of the sets parameter individually or for taking all together the Subcascades object is filtered for cascades that are made up of the same classes, a subset of classes (ordered = FALSE) or a cascade, part of a cascade (ordered = TRUE), resulting in a set of cascades that might contain less classes. The parameter neighbourhood is not considered.

If direction is set to 'super' the Subcascades object is filtered for a superset of the given cascades. If the type parameter is set to 'any' each cascade is considered individually otherwise all cascades are used as reference for filtering. A superset are cascades that are made up of (parts) of the given cascades (ordered = TRUE) or classes (ordered=FALSE) but can contain also more classes. Depending on the parameter neighborhood only classes defined in the sets parameter are allowed as neighbors ('direct'). If neighborhood is set to 'indirect' the filtering is less strict and the direct neighborhood defined in sets is not considered.

Value

A Subcascades object comprising the evaluated cascades and their performances. The Subcascades object is made up of a list of matrices. Each matrix comprises the evaluation results of cascades of a specific length and is sorted row-wise according to the achieved minimal classwise sensitivities of the cascades (decreasing). The rownames show the class order by a character string of type '1>2>3' and the entries the sensitivity for each position of the cascade.

See Also

dropSize, keepSize, keepSets, dropThreshold, keepThreshold

Examples

library(TunePareto)
data(esl)
data = esl$data
labels = esl$labels
foldList = generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap = predictionMap(data, labels, foldList = foldList, 
                       classifier = tunePareto.svm(), kernel='linear')
# generate Subcascades object
subc = subcascades(predMap,thresh=0.7)

#define sets
set1 = list(c(1,2,3),c(2,3,4))
set2 = c('1>2>3','2>3>4')

# filter for the subset cascades that contain either the classes 
# {1,2,3} or {2,3,4} independent of the order, but neighbored
dropSets(subc, sets = set1, direction = 'sub', 
         ordered = FALSE, neighborhood = 'direct')
dropSets(subc, sets = set2, direction = 'sub', 
         ordered = FALSE, neighborhood = 'direct')

# filter for the superset cascades that contain either the classes 
# {1,2,3} or {2,3,4} independent of the order, but neighbored
dropSets(subc, sets = set1, direction = 'super', 
         ordered = FALSE, neighborhood = 'direct')
dropSets(subc, sets = set2, direction = 'super', 
         ordered = FALSE, neighborhood = 'direct')

# filter for the superset cascades that contain both the classes 
# {1,2,3} and {2,3,4} in exactly the given order, but allowing 
# for other classes inbetween
dropSets(subc, sets = set1, direction = 'super', 
        ordered = TRUE, neighborhood = 'indirect', type = 'all')
dropSets(subc, sets = set2, direction = 'super',
        ordered = TRUE, neighborhood = 'indirect', type = 'all')
         
# filter for the exact cascades
# sets can be a numeric list
result <- dropSets(subc, list(c(0,1,2),c(2,3,4,1)),
                   direction = 'exact', ordered=TRUE)
unlist(t(lapply(result,rownames)))
# or sets can be a character vector
result <- dropSets(subc, c('0>1>2','2>3>4>1'),
                   direction = 'exact',ordered=TRUE)
unlist(t(lapply(result,rownames)))         

Filters for size

Description

Filters out the Subcascades object for the given sizes.

Usage

dropSize(subcascades = NULL, size = NA)

Arguments

Value

A Subcascades object as in subcascades, that does not include cascades of the specific lengths that hve been filtered.

See Also

keepSize, dropSets, keepSets, dropThreshold, keepThreshold

Examples

library(TunePareto)
data(esl)
data = esl$data
labels = esl$labels
foldList = generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap = predictionMap(data, labels, foldList = foldList, 
                       classifier = tunePareto.svm(), kernel='linear')
# generate Subcascades object
subc = subcascades(predMap,thresh=0.7)

# filters out cascades that have a length of 3
dropSize(subc,size=3)
# filters out cascades that have a length of 3 or 4
dropSize(subc, size=c(3,4))

Exclude Cascades Based on Threshold

Description

Filters out all cascades that match the comparison with a minimal classwise sensitivity threshold.

Usage

dropThreshold(subcascades = NULL, comparison = ">=", thresh = 0)

Arguments

Value

A Subcascades object comprising the evaluated cascades and their performances. The Subcascades object is made up of a list of matrices. Each matrix comprises the evaluation results of cascades of a specific length and is sorted row-wise according to the achieved minimal classwise sensitivities of the cascades (decreasing). The rownames show the class order by a character string of type '1>2>3' and the entries the sensitivity for each position of the cascade.

See Also

dropSize, keepSize, dropSets, keepSets, keepThreshold

Examples

library(TunePareto)
data(esl)
data = esl$data
labels = esl$labels
foldList = generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap = predictionMap(data, labels, foldList = foldList, 
                       classifier = tunePareto.svm(), kernel='linear')
# generate Subcascades object
subc = subcascades(predMap,thresh=0.5)

# filters for cascades that 
# 1. have a minimal classwise sensitivity >= 0.6
dropThreshold(subc,thresh=0.6)
# 2. have a minimal classwise sensitivity <= 0.6
dropThreshold(subc, comparison = '<=', thresh=0.6)

ESL Dataset

Description

The original ESL data set contains 488 profiles of applicants for certain industrial jobs. Expert psychologists of a recruiting company, based upon psychometric test results and interviews with the candidates, determined the values of the input attributes. The output is the an overall score corresponding to the degree of fitness of the candidate to this type of job. Here classes 1-3 are merged to one group and all classes >6 are merged to one group. Furtheremore, the data is relabeled to start at 0 and only 20 samples per class are included.

Usage

data(esl)

Format

An object of class list of length 2.

Source

Weka dataset (datasets-arie_ben_david.tar.gz, 11,348 Bytes)

References

ordinal, real-world datasets donated by Dr. Arie Ben David (Holon Inst. of Technology/Israel) Donor: Arie Ben David MIS, Dept. of Technology Management Holon Academic Inst. of Technology 52 Golomb St. Holon 58102 Israel abendav@hait.ac.il Owner: Yoav Ganzah Business Administration School Tel Aviv Univerity Ramat Aviv 69978 Israel

Examples

data(esl)

ESL Dataset

Description

The ESL data set contains 488 profiles of applicants for certain industrial jobs. Expert psychologists of a recruiting company, based upon psychometric test results and interviews with the candidates, determined the values of the input attributes. The output is the an overall score corresponding to the degree of fitness of the candidate to this type of job.

Usage

data(esl_org)

Format

An object of class data.frame with 488 rows and 5 columns.

Source

Weka dataset (datasets-arie_ben_david.tar.gz, 11,348 Bytes)

References

ordinal, real-world datasets donated by Dr. Arie Ben David (Holon Inst. of Technology/Israel) Donor: Arie Ben David MIS, Dept. of Technology Management Holon Academic Inst. of Technology 52 Golomb St. Holon 58102 Israel abendav@hait.ac.il Owner: Yoav Ganzah Business Administration School Tel Aviv Univerity Ramat Aviv 69978 Israel

Examples

data(esl_org)

Coerce to a Groupwise object

Description

Converts from a Subcascades object to a Groupwise object.

Usage

groupwise(subcascades = NULL, maxCl = 50)

Arguments

Details

This function re-sorts the Subcascades object in a way that the cascades made up of the same classes are grouped.

Value

A Groupwise object, which comprises a two-leveled list. The first level collects cascades of the same size. The second level contains a list of unique class combinations, labelled as a character string with '-' separating the different classes. For each unique set of class combinations the corresponding orders and their performance are given as a matrix, where each row contains a cascade, given as a character string of type '1>2>3', and the columns the sensitivity for the class at the corresponding position. Each matrix is sorted row-wise according to the achieved minimal classwise sensitivites of the cascades (decreasing).

See Also

as.subcascades, summary.Groupwise, print.Groupwise, plot.Groupwise, as.edgedataframe

Examples

library(TunePareto)
data(esl)
data = esl$data
labels = esl$labels
foldList = generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap = predictionMap(data, labels, foldList = foldList, 
                       classifier = tunePareto.svm(), kernel='linear')
# generate Subcascades object
subcascades = subcascades(predMap,thresh=0.7)

#create a Groupwise object
groupwise = groupwise(subcascades)

Filter Subcascades

Description

keepSets filters specific cascade sets.

Usage

keepSets(
  subcascades = NULL,
  sets = NULL,
  direction = "sub",
  ordered = FALSE,
  neighborhood = "direct",
  type = "any"
)

Arguments

Details

This function allows for filtering the Subcascades object.

If direction is set to 'exact' the parameter neighborhood is ignored and the parameter type is used as its default and cannot be changed. There has to be an exact match between the classes of the sets parameter cascades and the cascade of the Subcascades object, so not more but also not less classes are allowed. If the ordered parameter is set to TRUE also the order of the classes within the sets parameter and within the Subcascades object has be be exactly the same.

If direction is set to 'sub' the Subcascades object is filtered for subsets of the given cascades. If the type parameter is set to 'any' each cascade is considered individually otherwise all cascades are used as reference for filtering. This means that either for each cascade of the sets parameter individually or for taking all together the Subcascades object is filtered for cascades that are made up of the same classes, a subset of classes (ordered = FALSE) or a cascade, part of a cascade (ordered = TRUE), resulting in a set of cascades that might contain less classes. The parameter neighbourhood is not considered.

If direction is set to 'super' the Subcascades object is filtered for a superset of the given cascades. If the type parameter is set to 'any' each cascade is considered individually otherwise all cascades are used as reference for filtering. A superset are cascades that are made up of (parts) of the given cascades (ordered = TRUE) or classes (ordered=FALSE) but can contain also more classes. Depending on the parameter neighborhood only classes defined in the sets parameter are allowed as neighbors ('direct'). If neighborhood is set to 'indirect' the filtering is less strict and the direct neighborhood defined in sets is not considered.

Value

A Subcascades object comprising the evaluated cascades and their performances. The Subcascades object is made up of a list of matrices. Each matrix comprises the evaluation results of cascades of a specific length and is sorted row-wise according to the achieved minimal classwise sensitivities of the cascades (decreasing). The rownames show the class order by a character string of type '1>2>3' and the entries the sensitivity for each position of the cascade.

See Also

dropSize, keepSize, dropSets, dropThreshold, keepThreshold

Examples

library(TunePareto)
data(esl)
data = esl$data
labels = esl$labels
foldList = generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap = predictionMap(data, labels, foldList = foldList, 
                       classifier = tunePareto.svm(), kernel='linear')
# generate Subcascades object
subc = subcascades(predMap,thresh=0.7)

#define sets
set1 = list(c(1,2,3),c(2,3,4))
set2 = c('1>2>3','2>3>4')

# filter for the subset cascades that contain either the classes 
# {1,2,3} or {2,3,4} independent of the order, but neighbored
keepSets(subc, sets = set1, direction = 'sub', 
         ordered = FALSE, neighborhood = 'direct')
keepSets(subc, sets = set2, direction = 'sub', 
         ordered = FALSE, neighborhood = 'direct')

# filter for the superset cascades that contain either the classes 
# {1,2,3} or {2,3,4} independent of the order, but neighbored
keepSets(subc, sets = set1, direction = 'super', 
         ordered = FALSE, neighborhood = 'direct')
keepSets(subc, sets = set2, direction = 'super', 
         ordered = FALSE, neighborhood = 'direct')

# filter for the superset cascades that contain both the classes 
# {1,2,3} and {2,3,4} in exactly the given order, but allowing 
# for other classes inbetween
keepSets(subc, sets = set1, direction = 'super', 
        ordered = TRUE, neighborhood = 'indirect', type = 'all')
keepSets(subc, sets = set2, direction = 'super',
        ordered = TRUE, neighborhood = 'indirect', type = 'all')
         
# filter for the exact cascades
# sets can be a numeric list
result <- keepSets(subc, list(c(0,1,2),c(2,3,4,1)),
                   direction = 'exact', ordered=TRUE)
unlist(t(lapply(result,rownames)))
# or sets can be a character vector
result <- keepSets(subc, c('0>1>2','2>3>4>1'),
                   direction = 'exact',ordered=TRUE)
unlist(t(lapply(result,rownames)))         
 

Filter for Size

Description

Filters the Subcascades object for cascades of the given sizes.

Usage

keepSize(subcascades = NULL, size = NA)

Arguments

Value

A Subcascades object as in subcascades, that only includes cascades of specific lengths.

See Also

dropSize, dropSets, keepSets, dropThreshold, keepThreshold

Examples

library(TunePareto)
data(esl)
data = esl$data
labels = esl$labels
foldList = generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap = predictionMap(data, labels, foldList = foldList, 
                       classifier = tunePareto.svm(), kernel='linear')
# generate Subcascades object
subc = subcascades(predMap,thresh=0.7)

# filters for cascades that have a length of 3
keepSize(subc,size=3)
# filters for cascades that have a length of 3 or 4
keepSize(subc, size=c(3,4))

Filters for Threshold

Description

Filters for all cascades that match the comparison with a minimal classwise sensitivity threshold.

Usage

keepThreshold(subcascades = NULL, thresh = 0, comparison = ">=")

Arguments

Value

A Subcascades object comprising the evaluated cascades and their performances. The Subcascades object is made up of a list of matrices. Each matrix comprises the evaluation results of cascades of a specific length and is sorted row-wise according to the achieved minimal classwise sensitivities of the cascades (decreasing). The rownames show the class order by a character string of type '1>2>3' and the entries the sensitivity for each position of the cascade.

See Also

dropSize, keepSize, dropSets, keepSets, dropThreshold

Examples

library(TunePareto)
data(esl)
data = esl$data
labels = esl$labels
foldList = generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap = predictionMap(data, labels, foldList = foldList, 
                       classifier = tunePareto.svm(), kernel='linear')
# generate Subcascades object
subc = subcascades(predMap,thresh=0.5)

# filters for cascades that 
# 1. have a minimal classwise sensitivity >= 0.6
keepThreshold(subc,thresh=0.6)
# 2. have a minimal classwise sensitivity <= 0.6
keepThreshold(subc, comparison = '<=', thresh=0.6)

Merge Subcascades

Description

mergeSubcascades adds the cascades from subcascades2 to the subcascades1, that have not been part of subcascades1.

Usage

mergeSubcascades(subcascades1 = NULL, subcascades2 = NULL)

Arguments

Value

A Subcascades object comprising the evaluated cascades and their performances. The Subcascades object is made up of a list of matrices. Each matrix comprises the evaluation results of cascades of a specific length and is sorted row-wise according to the achieved minimal classwise sensitivities of the cascades (decreasing). The rownames show the class order by a character string of type '1>2>3' and the entries the sensitivity for each position of the cascade.

Examples

library(TunePareto)
data(esl)
data = esl$data
labels = esl$labels
foldList = generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap = predictionMap(data, labels, foldList = foldList, 
                       classifier = tunePareto.svm(), kernel='linear')

# make two Subcascades objects
subc1 = subcascades(predMap, size = c(3,4), thresh = 0.6)
subc2 = subcascades(predMap, size = c(4), thresh = 0.5)
# add the cascades of subcascades2 to subcascades1
mergeSubcascades(subc1, subc2)

non-ordinal data

Description

It comprises a list of two elements: A data matrix ("data") and a label vector ("labels"). The samples are given in the rows and the features in the columns of the matrix. The sampled dataset comprises 5 classes with 40 samples, each. It is embedded in a two dimensional feature space and contains 5 classes with no ordinal sequence.

Usage

data(nonOrdinalData)

Format

An object of class list of length 2.

Examples

data(nonOrdinalData)

ordinal data

Description

It comprises a list of two elements: A data matrix ("data") and a label vector ("labels"). The samples are given in the rows and the features in the columns of the matrix. The sampled dataset comprises 5 classes with 40 samples, each. It is embedded in a two dimensional feature space and contains 1 ordinal sequence with 5 classes.

Usage

data(ordinalData)

Format

An object of class list of length 2.

Examples

data(ordinalData)

Base Classifier Performance Heatmap

Description

Plots a heatmap that shows base classifier performance.

Usage

## S3 method for class 'Conf'
plot(
  x = NULL,
  classNames = NULL,
  onlySens = FALSE,
  symmetric = FALSE,
  main = "summary of base classifiers",
  xlab = "classes",
  ylab = "base classifiers",
  digits = 3,
  ignore = 0,
  first.colors = c("#f5f5f5", "#b2182b"),
  second.colors = c("#f5f5f5", "#2166ac"),
  other.colors = c("#f5f5f5", "#1b7837"),
  na.color = c("yellow"),
  las = 1,
  srt = 30,
  color.key = FALSE,
  cex = 1,
  cex.lab = 1,
  ...
)

Arguments

Details

This function plots a heatmap of the base classifier performance.

Value

No return value, called to generate the heatmap plot.

See Also

conf, plot.Subcascades, plot.ConfusionTable

Examples

library(TunePareto)
data(esl)
data <- esl$data
labels <- esl$labels
foldList <- generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap <- predictionMap(data, labels, foldList = foldList, 
                        classifier = tunePareto.svm(), kernel='linear')
# generate Subcascades object
conf <- conf(predMap)

plot(conf, onlySens=TRUE, symmetric=TRUE)

Extended Confusion Table Plot

Description

Plots an extended confusion table.

Usage

## S3 method for class 'ConfusionTable'
plot(
  x = NULL,
  classNames = NULL,
  main = "extended confusion table",
  xlab = "original class labels",
  ylab = "predicted class labels",
  cascLab = "inner classes",
  otherLab = "outer classes",
  digits = 3,
  ignore = 0,
  casc.colors = c("#f5f5f5", "#8c510a"),
  other.colors = c("#f5f5f5", "#01665e"),
  colSep = "#b2182b",
  las = 1,
  color.key = TRUE,
  cex = 1,
  cex.lab = 1,
  ...
)

Arguments

Value

No return value, called to generate the confusion table plot.

See Also

confusionTable, plot.Subcascades, plot.Conf

Examples

library(TunePareto)
data(esl)
data <- esl$data
labels <- esl$labels
foldList <- generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap <- predictionMap(data, labels, foldList = foldList, 
                        classifier = tunePareto.svm(), kernel='linear')
# generate Subcascades object
conf.table <- confusionTable(predMap,cascade='0>1>3>4',other.classes = 'all')

plot(conf.table)

Heatmap of a Groupwise Object

Description

Plots a heatmap, that shows the performance of cascades of a Groupwise object.

Usage

## S3 method for class 'Groupwise'
plot(
  x = NULL,
  class.sort = "",
  row.sort = "sens",
  main = "groupwise",
  xlab = "classes",
  ylab = "cascades",
  digits = 3,
  ignore = 0,
  casc.colors = c("#f5f5f5", "#01665e"),
  na.color = c("#f5f5f5"),
  color.key = TRUE,
  las = 1,
  cex = 1,
  cex.lab = 1,
  srt = 30,
  ...
)

Arguments

Details

This function plots a heatmap with the cascades of the Groupwise object in the rows and all classes present in any of the cascades in the columns. The colors indicate whether a given class is present in the corresponding cascade and with which sensitivity. Internally converts the Groupwise object to a Subcascades object and plots the corresponding heatmap.

Value

No return value, called to generate the heatmap plot of the Groupwise Object.

See Also

groupwise, subcascades, plot.Subcascades

Examples

library(TunePareto)
data(esl)
data <- esl$data
labels <- esl$labels
foldList <- generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap <- predictionMap(data, labels, foldList = foldList, 
                        classifier = tunePareto.svm(), kernel='linear')
# generate Subcascades object
subc <- subcascades(predMap,thresh=0.7,size=c(3,4))
groupwise <- groupwise(subc)

plot(groupwise,row.sort='max')

Heatmap of a PredictionMap Object

Description

Plots a heatmap, that shows the predictions of a PredictionMap object and the real labels in a cross-validation or reclassification experiment.

Usage

## S3 method for class 'PredictionMap'
plot(
  x = NULL,
  xlab = "samples",
  ylab = "base classifiers",
  main = "Prediction map",
  las = 1,
  srt = 30,
  cex = 1,
  cex.lab = 1,
  label.colors = NULL,
  plot.sampleIDs = FALSE,
  plot.cv.runs = TRUE,
  plot.class.labels = TRUE,
  ...
)

Arguments

Details

This function plots a heatmap with color-decoded predictions made by the specified classifier. Here, the rows indicate the different binary base classifiers and the columns the samples in the specified resampling experiment (reclassification or cross-validation). Labels are visualized in the top row and decoded color-wise.

Value

No return value, called to a heatmap plot of the predictionMap Object.

See Also

predictionMap

Examples

library(TunePareto)
data(esl)
data <- esl$data
labels <- esl$labels
foldList <- generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap <- predictionMap(data, labels, foldList = foldList, 
                        classifier = tunePareto.svm(), kernel='linear')

plot(predMap)

Heatmap of a Subcascades Object

Description

Plots a heatmap, that shows the performance of cascades of a Subcascades object.

Usage

## S3 method for class 'Subcascades'
plot(
  x = NULL,
  classNames = NULL,
  class.sort = "",
  row.sort = "",
  main = "subcascades",
  xlab = "classes",
  ylab = "cascades",
  digits = 3,
  ignore = 0,
  casc.colors = c("#f5f5f5", "#01665e"),
  na.color = c("#f5f5f5"),
  color.key = TRUE,
  las = 1,
  cex = 1,
  cex.lab = 1,
  srt = 30,
  ...
)

Arguments

Details

This function plots a heatmap with the cascades of the Subcascades object in the rows and all classes present in any of the cascades in the columns. The colors indicate whether a given class is present in the corresponding cascade and with which sensitivity.

Value

No return value, called to generate the heatmap plot of the subcascades Object.

See Also

subcascades, plot.Conf, plot.ConfusionTable

Examples

library(TunePareto)
data(esl)
data <- esl$data
labels <- esl$labels
foldList <- generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap <- predictionMap(data, labels, foldList = foldList, 
                        classifier = tunePareto.svm(), kernel='linear')
# generate Subcascades object
subc <- subcascades(predMap,thresh=0.7,size=c(3,4))

plot(subc,row.sort='max')

Base Classifier Performance Heatmap

Description

Plots a heatmap that shows base classifier performance.

Usage

plotBaseClassifier(
  x = NULL,
  classNames = NULL,
  symmetric = FALSE,
  main = "summary of base classifiers",
  xlab = "second class",
  ylab = "first class",
  digits = 3,
  ignore = 0,
  first.colors = c("#f5f5f5", "#8c510a"),
  second.colors = c("#f5f5f5", "#01665e"),
  na.color = c("yellow"),
  las = 1,
  color.key = TRUE,
  cex = 1,
  cex.lab = 1,
  ...
)

Arguments

Details

This function plots a heatmap of the base classifier performance.

Value

No return value, called to generate the heatmap plot.

Base Classifier Performance Heatmap

Description

Plots a heatmap that shows base classifier performance.

Usage

plotConf(
  x = NULL,
  classNames = NULL,
  main = "summary of base classifiers",
  xlab = "classes",
  ylab = "base classifiers",
  digits = 3,
  ignore = 0,
  first.colors = c("#f5f5f5", "#b2182b"),
  second.colors = c("#f5f5f5", "#2166ac"),
  other.colors = c("#f5f5f5", "#1b7837"),
  las = 1,
  srt = 30,
  color.key = FALSE,
  cex = 1,
  cex.lab = 1,
  ...
)

Arguments

Details

This function plots a heatmap of the base classifier performance.

Value

No return value, called to generate the heatmap plot.

See Also

conf, plot.Subcascades, plot.ConfusionTable

Construction of a Prediction Map Object

Description

Makes a PredictionMap object for the given data.

Usage

predictionMap(
  data = NULL,
  labels = NULL,
  foldList = NULL,
  parallel = FALSE,
  classifier = NULL,
  ...
)

Arguments

Details

Using a reclassification or a cross-validation set up this function performs a pairwise training for all class combinations and evaluates all samples using the trained classifiers. This means that for each run, fold and binary classifier the predicted class for each sample is calculated.

Value

A PredictionMap object. It is made up of a list of two matrices, which are called meta and pred. Both matrices provide information for individual samples column-wise. The meta information in meta connects the values in the pred-matrix to a specific fold, run, sample and contains the original label. The rownames of the pred-matrix (e.g. [0vs1]) show the classes of the binary base classifier. The elements are the prediction result of a specific training. The rows that correspond to base classifiers that would separate the same class consists of -1. Those rows are not used within the analysis.

See Also

summary.PredictionMap, print.PredictionMap, plot.PredictionMap

Examples

library(TunePareto)
data(esl)
data = esl$data
labels = esl$labels
foldList = generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
                          
# svm with linear kernel
predMap = predictionMap(data, labels, foldList = foldList, 
                       classifier = tunePareto.svm(), kernel='linear')

# knn with k = 3
predMap = predictionMap(data, labels, foldList = foldList, 
                       classifier = tunePareto.knn(), k = 3)
# randomForest
predMap = predictionMap(data, labels, foldList = foldList, 
                       classifier = tunePareto.randomForest())
                           

Prints all details about a Conf object.

Description

Prints all details about a Conf object.

Usage

## S3 method for class 'Conf'
print(x, printfC = TRUE, printsC = TRUE, ...)

Arguments

Details

Prints all details about a given Conf object. A Conf object provides the conditional prediction rates of all base classifiers applied to all classes.

Examples

library(TunePareto)
data(esl)
data = esl$data
labels = esl$labels
foldList = generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap = predictionMap(data, labels, foldList = foldList, 
                       classifier = tunePareto.svm(), kernel='linear')

conf = conf(predMap)

print(conf)

Prints all details about a ConfusionTable object.

Description

Prints all details about a ConfusionTable object.

Usage

## S3 method for class 'ConfusionTable'
print(x, ...)

Arguments

Details

Prints all conditional prediction rates for the specified cascade.

Examples

library(TunePareto)
data(esl)
data = esl$data
labels = esl$labels
foldList = generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap = predictionMap(data, labels, foldList = foldList, 
                       classifier = tunePareto.svm(), kernel='linear')

# Calculation of the confusion matrix for '0>2>3>4'.
confTable = confusionTable(predMap, cascade = '0>2>3>4')

print(confTable)

Prints all details about a Groupwise object.

Description

Prints all details about a Groupwise object.

Usage

## S3 method for class 'Groupwise'
print(x, printSizes = length(x), ...)

Arguments

Details

Prints all details about a given Groupwise object. A Groupwise object re-sorts a Subcascades object in a way that the cascades made up of the same classes are grouped. The printSizes parameter can be used to control the cascade sizes to be printed.

Examples

library(TunePareto)
data(esl)
data = esl$data
labels = esl$labels
foldList = generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap = predictionMap(data, labels, foldList = foldList, 
                       classifier = tunePareto.svm(), kernel='linear')

# generate Subcascades object
subc = subcascades(predMap,thresh=0.7)

#create a Groupwise object
groupwise = groupwise(subc)

print(groupwise, printSizes = 2)

Prints all details about a PredictionMap object.

Description

Prints all details about a PredictionMap object.

Usage

## S3 method for class 'PredictionMap'
print(x, showMeta = TRUE, showPred = TRUE, ...)

Arguments

Details

Prints all details about a given PredictionMap object and information about the meta information as well as the predictions.

Examples

library(TunePareto)
data(esl)
data = esl$data
labels = esl$labels
foldList = generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
                          
# svm with linear kernel
predMap = predictionMap(data, labels, foldList = foldList, 
                       classifier = tunePareto.svm(), kernel='linear')
                           
print(predMap, showMeta=TRUE, showPred=TRUE)

Prints all details about a Subcascades object.

Description

Prints all details about a Subcascades object.

Usage

## S3 method for class 'Subcascades'
print(x, printSizes = length(x), ...)

Arguments

Details

Prints all details about a given Subcascades object.

Examples

library(TunePareto)
data(esl)
data = esl$data
labels = esl$labels
foldList = generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap = predictionMap(data, labels, foldList = foldList, 
                       classifier = tunePareto.svm(),  kernel='linear')

# use default parameter settings 
# -> returns cascades of all lengths that show a minimal classwise sensitivity >0.
subc = subcascades(predMap)

# print subcascades for the largest 2 sizes
print(subc, printSize=2)

screening data

Description

It comprises a list of two elements: A data matrix ("data") and a label vector ("labels"). The samples are given in the rows and the features in the columns of the matrix. The sampled dataset comprises 6 classes with 40 samples, each. It is embedded in a two dimensional feature space and contains 2 ordinal sequences with 5 classes.

Usage

data(screeningData)

Format

An object of class list of length 2.

Examples

data(screeningData)

Subcascades Evaluation

Description

Subcascades returns all cascades found within the data or evaluates a set of specific cascades.

Usage

subcascades(
  predictionMap = NULL,
  sets = NULL,
  thresh = 0,
  size = NA,
  numSol = 1000
)

Arguments

Details

This function can either be used to evaluate the performance of a specific cascade, a set of cascades or to filter out the set of cascades of a specific size and passing a given threshold. If the sets-variable is given no size can be set.

Value

A Subcascades object comprising the evaluated cascades and their performances. The Subcascades object is made up of a list of matrices. Each matrix comprises the evaluation results of cascades of a specific length and is sorted row-wise according to the achieved minimal classwise sensitivities of the cascades (decreasing). The rownames show the class order by a character string of type '1>2>3' and the entries the sensitivity for each position of the cascade.

See Also

print.Subcascades, plot.Subcascades, summary.Subcascades

Examples

library(TunePareto)
data(esl)
data = esl$data
labels = esl$labels
foldList = generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap = predictionMap(data, labels, foldList = foldList, 
                       classifier = tunePareto.svm(),  kernel='linear')

# use default parameter settings 
# -> returns cascades of all lengths that show a minimal classwise sensitivity >0.
subc = subcascades(predMap)
# change the threshold 
# -> returns cascades of all lengths that show a minimal classwise sensitivity >=0.6.
subc = subcascades(predMap, thresh=0.6)
# search only for cascades of length 2 and 4 
# -> returns cascades of length 2 and 4 that show a minimal classwise sensitivity >=0.6.
subc = subcascades(predMap, thresh=0.6, size=c(2,4))
# evaluates the performance of the cascade '0>1>2>3>4'.
subc = subcascades(predMap, sets = c('0>1>2>3>4'))

Summary of conf

Description

summary.Conf returns a summarizing overview of a Conf object.

Usage

## S3 method for class 'Conf'
summary(object = NULL, ...)

Arguments

Details

This function gives an overview of the Conf object. The overall number of classes, the highest and lowest sensitivities of the first class as well as the highest and lowest performance of the second class are recorded.

See Also

conf, summary.PredictionMap, summary.Subcascades, summary.Groupwise, summary.ConfusionTable

Examples

library(TunePareto)
data(esl)
data = esl$data
labels = esl$labels
foldList = generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap = predictionMap(data, labels, foldList = foldList, 
                       classifier = tunePareto.svm(), kernel='linear')

conf = conf(predMap)

summary(conf)

Summary of confusion tables

Description

summary.ConfusionTable returns a summarizing overview.

Usage

## S3 method for class 'ConfusionTable'
summary(object = NULL, ...)

Arguments

Details

This function gives an overview of the characteristics of the ConfusionTable object. The cascade with the corresponding mean accuracy are given.

See Also

confusionTable, summary.PredictionMap, summary.Subcascades, summary.Groupwise, summary.Conf

Examples

library(TunePareto)
data(esl)
data = esl$data
labels = esl$labels
foldList = generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap = predictionMap(data, labels, foldList = foldList, 
                       classifier = tunePareto.svm(), kernel='linear')

# Calculation of the confusion matrix for '0>2>3>4'.
confusionTable = confusionTable(predMap, cascade = '0>2>3>4')

summary(confusionTable)

Overview Class Groups

Description

summary.Groupwise returns a summarizing overview. For each length the number of permutations consisting of the same set of classes is given.

Usage

## S3 method for class 'Groupwise'
summary(object = NULL, ...)

Arguments

Details

This function gives an overview of the characteristics of the Groupwise object. The number of permutations per size is given. A permutation means that the corresponding cascades contain the same classes but with different order.

See Also

groupwise, summary.PredictionMap, summary.Subcascades, summary.Conf, summary.ConfusionTable

Examples

library(TunePareto)
data(esl)
data = esl$data
labels = esl$labels
foldList = generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap = predictionMap(data, labels, foldList = foldList, 
                       classifier = tunePareto.svm(), kernel='linear')
# generate Subcascades object
subc = subcascades(predMap,thresh=0.7)
groupwise = groupwise(subc,maxCl=50)

summary(groupwise)

Summary of prediction maps

Description

summary.PredictionMap returns a summarizing overview of a PredictionMap object.

Usage

## S3 method for class 'PredictionMap'
summary(object = NULL, ...)

Arguments

Details

This function gives an overview of the PredictionMap object. A short summary about the utilized data and labels is given as well as the number of runs and folds of the cross-validation. The summary also includes if the prediction map was generated in parallel and the name of the utilized TunePareto classifier as well as its specified parameters.

See Also

predictionMap, summary.Subcascades, summary.Groupwise, summary.Conf, summary.ConfusionTable

Examples

library(TunePareto)
data(esl)
data = esl$data
labels = esl$labels
foldList = generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap = predictionMap(data, labels, foldList = foldList, 
                       classifier = tunePareto.svm(), kernel='linear')

summary(predMap)

Summary Subcascades Characteristics

Description

Generates a general overview of the characteristics of the Subcascades object.

Usage

## S3 method for class 'Subcascades'
summary(object = NULL, includeClassSummary = TRUE, digits = 3, ...)

Arguments

Details

This function gives a general overview of characteristics of the Subcascades object, like number of cascades or maximal cascade length.

See Also

subcascades, summary.PredictionMap, summary.Groupwise, summary.Conf, summary.ConfusionTable

Examples

library(TunePareto)
data(esl)
data <- esl$data
labels <- esl$labels
foldList <- generateCVRuns(labels  = labels,
                          ntimes      = 2,
                          nfold       = 2,
                          leaveOneOut = FALSE,
                          stratified  = TRUE)
predMap <- predictionMap(data, labels, foldList = foldList, 
                        classifier = tunePareto.svm(), kernel='linear')
# generate Subcascades object
subc <- subcascades(predMap,thresh=0.7,numSol=10000)

summary(subc)

Occurrence of Classes by Size

Description

summaryClasses returns the occurrence of classes by size

Usage

summaryClasses(subcascades = NULL)

Arguments

Details

This function gives an overview of the classes of the Subcascades object. For each length in the Subcascades object the occurence of classes is given.

Value

A matrix summarizing the overview characteristics of the Subcascades object.

Overview Subcascades

Description

summarySubcascades returns a summarizing overview. For each length the number of cascades and the minimal class-wise sensitivity is given.

Usage

summarySubcascades(subcascades = NULL)

Arguments

Details

This function gives an overview of the Subcascades object. For each length in the Subcascades object the number of cascades of that length, as well as their minimal classwise sensitivity is given.

Value

A matrix summarizing the overview characteristics of the Subcascades object. For each size (rows) the number of cascades within the Subcascades object (number) and the minimal classwise sensitivity (min.class.sens) are given.

Ordinal Classifier Cascade Tune Pareto Object

Description

TunePareto wrapper for the ordinal classifier cascade.

Usage

tunePareto.occ(base.classifier)

Arguments

Details

The "tunePareto.occ" encapsulates the classifier of an ordinal classifier cascade. Additionally, to the parameters of the corresponding base classifier the "class.order" parameter can be provided. It is either a character vector, a numeric vector, or a vector representing a cascade of the following format '1>2>4'.

Value

Returns an object of class TuneParetoClassifier (see: TunePareto::tuneParetoClassifier). This can be passed to the function trainTuneParetoClassifier (see: TunePareto::trainTuneParetoClassifier).

Examples

library(TunePareto)
data(esl)
data = esl$data
labels = esl$labels
# train classifier
model <- trainTuneParetoClassifier( 
         classifier  = tunePareto.occ( base.classifier = tunePareto.svm()),
         trainData   = data,
         trainLabels = labels,
         class.order = as.character(c(4,3,1,0)),
         kernel      = "linear",
         cost        = 1)
         
# predict labels
prediction <- predict(object = model, newdata = data)