From Eigenvector Research Documentation Wiki
Jump to navigation Jump to search


Extract and display scores information from model.


a = plotscores(modl,options)
a = plotscores(modl,pred,options)
a = plotscores(scores,labels,classes)


Given a standard model structure, relevant scores information (e.g. labels) is collected and passed to plotgui for plotting.

If no output is requested, then plotscores initiates an interactive plotting utility to make scores plots. If an output is requested, no plots are made, and the output a is a dataset object containing the scores, labels, and other information that is needed for subsequent plotting.

This functionality is also accessible through the .plotscores method of EVRIModel Objects.


  • modl = a standard model structure. plotscores will extract the scores, labels, and and other information that is needed for plotting.
  • pred = a standard prediction structure (e.g. output from PCA call: pred = pca(x,modl,options);)
  • scores = a M by K scores matrix (class "double").

Optional Inputs

  • labels is a character or cell array with M rows containing sample labels
  • classes a vector with M integer elements of class identifiers.
  • Optional input options is discussed below.


  • a = a dataset containing scores and label information that can be passed to plotgui.


options = a structure array with the following fields:

  • plots: ['none' | 'final' | {'auto'} |], governs plotting behavior
    • 'auto' makes plots if no output is requested {default},
  • figure: [],governs where plots are made
    • when figure = [] plots are made in a new figure window {default}, this can also be a valid figure number (i.e. figure handle)
  • sct: [ 0 | {1} ], tells whether to plot cal (modl scores) with test (pred scores), sct = 1 plots original calibration data with prediction set {default = 1},
  • autoclass: [{0} | 1 ] when enabled (1), classes are automatically assigned to samples using the density-based scanning method (see dbscan). Assignment is done by locating samples which are close to each other in the multivariate score space OR, if sammon mapping has been done, in the reduced Sammon mapped space.
  • knnscoredistance: [ 3 ] governs the inclusion of KNN score distance metric in factor-based methods. If >0, this option defines the number of neighbors to use in calculating the KNN Score Distance (see knnscoredistance function). If zero, KNN Score Distance is omitted from the scores.
  • showerrorbars: [ 0 | {1} ] governs default display of error bars (when available). If 1, error bars are shown automatically. If 0, user must use checkbox to enable error bar display.
  • hiddenstats: {} list of statistics which should be hidden from scores plot right-click menu.
  • title: [ {'off'} | 'on' ] governs inclusion of title on figures and in output DataSet. When 'on' text description of content (including source name) will be included on plots and in .title{1} field of output.
  • reducedstats: [ {'none'} | 'only' | 'both' ] governs reporting of statistics as "reduced" (normalized to confidence limit) when possible. 'both' returns both reduced and regular stats. 'only' returns only reduced stats. 'none' returns only regular stats.

The default options can be retreived using: options = plotscores('options');.

See Also

analysis, knnscoredistance, mcr, modelstruct, modelviewer, mpca, pca, pcr, plotgui, ploteigen, plotscores_defaults, plotloads, pls