Skip to content
  • use of row and column labels were completely rewritten and simplified (also in plots)
  • functionality for using vector of numbers as column labels has been improved significantly
  • same works now for row labels (e.g. you can specify time or similar values as row labels)
  • both column and row values are now inherited when any transformations or models are used
  • you can also specify the values directly using properties rowValuesAll and colValuesAll
  • fixed several minor bugs and made small improvements mainly for plot methods
  • fixed several small bugs in model methods
Assets 3

@svkucheryavski svkucheryavski released this Jan 7, 2016 · 55 commits to master since this release

  • new behaviour for all statistical methods (mean(), std(), ...), now values are calculated for each column of a dataset
  • manual x-values now can be provided to gplot() (similar to plot())
  • if a vector with numbers is provided for column names, the numbers will be used as x-values for line plots
  • fixed a bug in classification plot which worked incorrectly if reference data is not provided
  • other small improvememnts to the plotting methods for mdadata objects
  • help and GitBook documentation were adjusted according to the changes
Assets 3

@svkucheryavski svkucheryavski released this Nov 23, 2015 · 64 commits to master since this release

  • new plots for PLS model: plotxloadings(), plotxyloadings() and plotweights()
  • small improvements for regression coefficients plot
  • methods scatter() and plot() now have a new parameter 'Groupby' for easy color grouping
  • method scatter() has a new parameter 'ShowContour' to draw a contour for a cluster of points
Assets 3

@svkucheryavski svkucheryavski released this Nov 19, 2015 · 68 commits to master since this release

  • fixed bug in predict() method for regression, which did not work if y references have not been provided
  • small improvements to the regcoeffs class
  • small improvements to the mdamlr class
  • fixed a bug in decomp, which lead to an error when mdaimage is used as a data source
  • NIPALS algorithm is implemented for PCA ('nipals')
  • The mdapca class got a new method — biplot()
Assets 3

@svkucheryavski svkucheryavski released this Nov 16, 2015 · 75 commits to master since this release

  • fixed bug in SIMCA which did not allow to use predict()
  • performance plots for classification do not require class number any more
  • overview plot for SIMCA and PCA now works correctly if only one component selected
  • bar and line plots for performance now works correctly if only one component is selected
Assets 2

@svkucheryavski svkucheryavski released this Nov 16, 2015 · 78 commits to master since this release

This release accumulates changes for v.0.1.0 and v.0.1.1

v.0.1.1

  • fixed bug in PLS-DA which did not allow to use predict()
  • fixed bug with wrong calculation of false positives

v. 0.1.0

  • SIMCA one-class classification is implemented
  • Multiplicative Scatter correction added to preprocessing
  • any math function (e.g. log, power, etc) can now be added to preprocessing object
  • several changes in names of variables and parameters (e.g. Q2 residuals -> Q residuals)
  • fixed a bug when changing parameter Alpha in PCA did not change the statistical limits
  • JK confidence intervals on regression coefficients plot are shown as lines if line plot is used
  • any ldecomp object (e.g. PCA or PLS results) has a new property residuals, E = X - TP'
  • plots for individual object and variable residuals are available (plotobjres, plotvarres)
  • small bug fixes and optimization
  • from this release, the release versions will have more sound structure: x.0.0. for major releases
    (significant changes in code), 0.x.0 for minor releases (new methods or functionality),
    0.0.x for bug fixes and small improvements
Assets 3
Nov 16, 2015
Merge branch 'develop'

@svkucheryavski svkucheryavski released this Oct 23, 2015 · 98 commits to master since this release

  • color grouping was improved for the case when number of unique values in colour grouping vector is relatively small (2-8)
Assets 3

@svkucheryavski svkucheryavski released this Oct 19, 2015 · 101 commits to master since this release

  • fixed compatibility issues with R2015b
  • PCA like model can be now obtained ICA algorithm
  • new preprocessing methods: "whitening", "norm" and "ref2abs"
  • bugs fixes and code improvements
  • documentation (HTML) has been moved to GitBook
  • old GUI tools were removed, new under developing
Assets 3

@svkucheryavski svkucheryavski released this May 21, 2015 · 107 commits to master since this release

  • fixed a bug lead to wrong factor levels when concatenate datasets
  • small bugs fixes and code improvements
Assets 2
You can’t perform that action at this time.