classify: Classification of the one dimensional points in a Pcp or Mlp...

In ClusterSignificance: The ClusterSignificance package provides tools to assess if class clusters in dimensionality reduced data representations have a separation different from permuted data

Description

Classification based on ROC params (TN TP FP FN).

Usage

## S4 method for signature 'ClassifiedPoints'
getData(x, n = NULL)

classify(x, ...)

## S4 method for signature 'Pcp'
classify(x, ...)

## S4 method for signature 'Mlp'
classify(x, ...)

## S4 method for signature 'ClassifiedPoints'
initialize(.Object, ..., scores,
  scores.points = scores.points, scores.index = scores.index, ROC, AUC,
  class.color)

## S4 method for signature 'ClassifiedPoints,missing'
plot(x, y, comparison = "all",
  class.color = NULL, ...)

## S4 method for signature 'ClassifiedPoints'
show(object)

Arguments

x	Pcp or Mlp Object for the function classify otherwise it is a ClassifiedPoints object
n	data to extract from ClassifiedPoints (NULL gives all)
...	additional arguments to pass on
.Object	internal object
scores	final scores
scores.points	sorted points
scores.index	index of sorted points
ROC	parameters (TN, TP, FN and FP)
AUC	area under the curve
class.color	user assigned group coloring scheme
y	default plot param, which should be set to NULL
comparison	Specify a comparison i.e. ("grp1 vs grp2") and plot only that comparison.
object	ClassifiedPoints Object

Details

Tests all possible discrimination lines and picks the one with highest score based on a score which is simply calculated by the formula (TP - FP) + (TN - FN).

The plot shows the distribution of scores for different discrimination lines. Each line is a separator that has a score for the separation of the two groups, and the height of the line marks the score for this separation.

Value

The classify function returns an object of class ClassifiedPoints

Author(s)

Jesper R. Gadin and Jason T. Serviss

Examples

#use demo data
data(pcpMatrix)
classes <- rownames(pcpMatrix)

#run function
prj <- pcp(pcpMatrix, classes)
cl <- classify(prj)

#getData accessor
getData(cl)

#getData accessor specific
getData(cl, "scores")

#plot result
plot(cl)

ClusterSignificance documentation built on Nov. 8, 2020, 5:28 p.m.