Release Notes Version 6 7

From Eigenvector Documentation Wiki
Jump to: navigation, search

Version 6.7 of PLS_Toolbox and Solo was released in March, 2012.

For general product information, see PLS_Toolbox Product Page. For information on Solo, see Solo Product Page.

This release was done in conjunction with MIA_Toolbox / Solo+MIA version 2.7

(back to Release Notes PLS Toolbox and Solo)

New Features

Analysis Window

  • Drag and Drop - Import files by simply dragging and dropping onto the Analysis window.
  • Confusion Table - Show cross-validated results when available.
  • Improved modelcache performance and string handling.
  • Updated toolbar icons.
  • Variance captured information now available for PLS, CLS, LWR, and MCR models. Information is also included report writer results.
  • X_hat (Data estimate) now available for test/prediction data.
  • Export models to ASCII-MOD file file format.
  • Selections in scores and loadings plots linked to Q / T^2 Contribution selections to make it easy to see what samples are being examined.
  • Add kNN Score Distance for MCR, Parafac, Purity, and CLS models.

Workspace Browser

  • Drag and Drop - Import files by simply dragging and dropping onto the Workspace Browser window. Also load, combine, and use data within Workspace Browser tools by dragging and dropping from any of the panes (Model cache/Demo data, Workspace, Analysis tools) or even into any open Analysis window. Improved highlighting when dropping into Analysis tools pane.
  • Easier to interpret demo data descriptions.

Import / Export

  • hjyreadr - Improved ability to install and manage ActiveX object.
  • xlsreadr - Add support for Apache POI java API for Windows documents for improved cross platform Excel file reading.

Plot Controls and Visualization Tools

  • Selectable plot themes (White, Black, or Grey backgrounds)
  • Custom Color-By values can now be loaded from (row/column) of a DataSet or matrix.
  • X-axis direction control using Plot Controls option 'axisautoinvert' (setting of -1 forces axis to be inverted)
  • Improved use of axis scales for multi-way data.
  • Copy/Paste of figure position feature added to FigBrowser menu (allows matching of figure sizes exactly)

DataSet Editor

  • Plot tab added to allow easy access to data plot.

Other Method Improvements

  • Classification methods - include "Class Pred Member - multiple" indicating when a sample is being assigned to more than one class.
  • Classification methods - "Strict Class Assignment" classification threshold can be adjusted using the 'strictthreshold' option (default = 50%).
  • Locally Weighted Regression (LWR) - More cross-validation results available.
  • iPLS/iPLSDA/iMLR - improved plots
  • PARAFAC - added "scaletype" option to give control over the type of normalization used for loadings (unit area, unit length (norm), maximum value)
  • MCR / ALS - added "normorder" option to give control over type of normalization used for loadings (unit area, unit length, maximum value)
  • SVM & SVMDA - improved automatic parameter selection.
  • Design Of Experients (DOE) Tools - updated DOE effects plot behavior, now displays all plots.

New Command-line Features and Functions

  • Full Support for Matlab R2012a

Command-line Tool Changes

  • Boxplot - Major overhaul:
    • Add many additional options to better match Mathworks' boxplot.
    • Better support for nice legends.
    • Modify defaults to match Mathworks boxplot.
    • Allow more color and symbol customization.
    • Improved help (making it easier to understand the limits calculations used.)
  • Options -allow throwing of error, warning, or command-line display when an option passed to a function contains invalid fields. Helps identify when you have mistyped an option name or used the wrong name. See Working With Function Options.
  • xclreadr/parsemixed - add "leadingdelim" option to allow ignoring leading delimiters (rather than replacing them with missing data.)
  • DataSet Object
    • Discontinue throwing of error when user attempts to index into non-existent but otherwise valid set. Instead simply return an empty (of appropriate type). This makes set indexing more friendly (the set doesn't exist, but it CAN exist, so returning empty value is more valid than throwing error).
    • Allow indexing by single index when ROW vector (e.g. x(1:5) when x is a row vector should return COLUMNS 1-5).
  • Windowfilter - significant speed improvement and added support for include field (variables).

Misc New Functions

classsummary - List class and axisscale distributions for a DataSet.
figuretheme - Resets a figure background and axes to a specified color.
windowfilter - Spectral filtering.