cancerclass (version 1.16.0)

plot: Plot Method for 'validation, nvalidation, prediction, predictor' Classes

Description

• class nvalidation: Plot of misclassification rates in dependence of the number of features that were used for classification. Total, class1 and class2 misclassification rates including confidence intervals can be plotted separately.
• class prediction: Plot methods for continuous predictions scores. Prediction scores are obtained by validation of a predictor in a test set. Four methods for assessment and visualization of predictor performance can be selected by the parameter type.
• class predictor: Among the three continuous prediction scores (z, zeta and ratio), zeta has the special property to be a linear combination of gene expression values. The plot method works only for the prediction score zeta and visualizes the contribution of each gene to the score.
• class validation: Plots the misclassification rate in dependence of the training set size. Total, class1 and class2 misclassification rates can be plotted including confidence intervals.

Usage

plot(x, y, ...)

Arguments

x

Object of class nvalidation.

y

missing

...

Further arguments directly passed to plot.

Methods

x

Object of class validation

y

missing

type

Three different kinds of plots can be generated: a xy-plot showing the misclassification rate in dependence of the training set size (type="xy"), a barplot showing the misclassification rates for each of the samples (type="samples"), a barplot showing how often (in %) a gene is included in a predictor (type="genes").

method

A character vector specifying the types of misclassification rates to be plotted. Possible types are the names of the classes and all for the total misclassification rate.

anno

Only relevant if type="genes": annotation of the features by array probes (anno="probe") or gene symbols (anno="symbol").

sig

Vector of numerical values corresponding to the method vector. The numerical values are equal to the confidence level, if equal to NULL, the corresponding confidence interval is not plotted.

xlog

A logical value. If TRUE, a logarithmic scale is used for the x-axis.

pos

Position of legend specified by a keyword from the list "bottomright", "bottom", "bottomleft", "left", "topleft", "top", "topright", "right" and "center".

ntrain

Only relevant if type="samples" or type="genes". Either results for predictors trained in training sets of different sample sizes (ntrain="all"), or results for predictors trained in training sets of the sample size specified by ntrain.

min.percent

Only relevant if type="genes". Theshold for selection of the genes that are plotted.

n

Only relevant if type="genes": Number of top genes that are plotted.

col

Only relevant if type="samples". Color of the boxes for samples of class1 and of class2.

ylim

Range of the y-axis.

cex.names

Only relevant if type="samples" or type="genes". Scaling factor for the labels of the x-axis.

col.curves

Only relevant if type="xy". A vector of strings corresponding to curves. Specifies the colors of the plot curves.

...

Further arguments directly passed to plot.

Methods

x

Object of class nvalidation

y

missing

type

Three different kinds of plots can be generated: a xy-plot showing the misclassification rate in dependence of the training set size (type="xy"), a barplot showing the misclassification rates for each of the samples (type="samples"), a barplot showing how often (in %) a gene is included in a predictor (type="genes").

method

A character vector specifying the types of misclassification rates to be plotted. Possible types are the names of the classes and all for the total misclassification rate.

anno

Only relevant if type="genes": annotation of the features by array probes or gene symbols.

sig

Vector of numerical values corresponding to the method vector. The numerical values are equal to the confidence level, if equal to NULL, the corresponding confidence interval is not plotted.

xlog

A logical value. If TRUE, a logarithmic scale is used for the x-axis.

pos

Position of legend specified by a keyword from the list "bottomright", "bottom", "bottomleft", "left", "topleft", "top", "topright", "right" and "center".

ngenes

Only relevant if type="samples" or type="genes". Either results for predictors including different number of genes (ngenes="all"), or results for predictors including the number of genes specified by the numeric value ngenes.

min.percent

Only relevant if type="genes". Theshold for selection of the genes that are plotted.

n

Only relevant if type="genes": Number of top genes that are plotted.

col

Only relevant if type="samples". Color of the boxes for samples of class1 and of class2.

ylim

Range of the y-axis.

cex.names

Only relevant if type="samples" or type="genes". Scaling factor for the labels of the x-axis.

...

Further arguments directly passed to plot.

Methods

x

Object of class prediction

y

missing

type

Four different kinds of plots can be generated: a histogram showing the distribution of the prediction score in class1 and class2 (type="histogram"), a xy-plot showing sensitivity, specificity, positive prediction value (PPV) and negative prediction value (NPV) in dependence on cutoffs for the prediction score (type="curves"), an ROC curve including calculation of the area under the curve (type="roc"), a barplot showing the prediction score for each of the samples (type="samples").

score

Specification of the prediction scores used for the plot: score="z", "zeta" or "ratio". If type="roc" a comparative analysis of two or three scores can be done.

breaks.dist

Distance of breaks.

ci

Only relevant, if type="roc". The method to calculate confidence intervals for sensitivity and specificity. Possibly values are "exact", "ac", "asymptotic", "wilson", "prop.test", "bayes", "logit", "cloglog", "probit", see R package binom for details.

col

Only relevant if type="samples". Numerical or character vector of length two specifying the color of symbols for correct and wrong classifications.

curves

Only relevant if type="curves". A vector of strings specifying the curves that are included into the plot. Can include "sensitivity", "specificity", "PPV" and "NPV".

col.curves

Only relevant if type="curves". A vector of strings corresponding to curves. Specifies the colors of the plot curves.

lty

Only relevant if type="roc". Numerical or character vector corresponding to the vector score. Specifies the line types used for the ROC plot.

npoints

Only relevant if type="logistic". Number of points to be plotted.

alpha

Only relevant if type="logistic". The probability of class membership is estimated by logistic regression. The parameter alpha specifies the confidence level for the confidence interval of this probability.

main

Title of the plot.

cex.names

Only relevant if type="samples". Scaling factor for the labels of the x-axis.

...

Further arguments directly passed to plot.

Methods

x

Object of class validation

y

missing

type

Currently only type="genes" is supported.

ngenes

Number of genes in the predictor.

dist

Character string specifying the method for calculation of the distance between test samples and the centroids. Possible values are "euclidean", "angle", "cor", "center".

anno

Annotation of the features by array probes (anno="probe") or gene symbols (anno="symbol").

ylab

Label of the y-axis.

main

Title of the plot.

...

Further arguments directly passed to plot.

See Also

validate, validation

Examples

### see: help(GOLUB);