

Changes in 5.0-3
================

Bug fixes:
----------
-the perf and predict functions have been updated. The prediction values are calculated based on the regression coefficients of Y onto the latent variables associated to X.

-scaling issues in perf/old-valid have been fixed

-one warning on the plotIndiv.rcc has been fixed.

-transition from valid() to perf() announced.

Changes in 5.0-2
================

New features:
-------------
- The valid function has been superseded by the perf function. Although similar in essence, few bugs have been fixed to estimate the performance of the sPLS and sPLS-DA models with no selection bias. A variable stability frequency has been added to the output. Functions spls.model and pls.model have been removed.

Bug fixes:
----------
-pls and spls function have been modified and harmonised w.r.t to scaling. Loading vectors a and b are now scaled to 1. Latent variables t and u are not scaled (following Table 21 of the Tenenhaus book - which is in French, sorry!).

-the argument abline.line has been set to FALSE by default in all plotIndiv functions.

- tune.multilevel for one factor has been fixed.

Changes in 5.0-1
================

New features:
-------------

New dependency to RGCCA package to enable integration of multiple matching data sets
- wrapping method wrapper.sgcca() and wrapper.rgcca() created
- S3 methods plotIndiv, plotVar, select.var, print for rgcca and sgcca added

Multilevel analysis
- cross validation enabled in function tune.multilevel for one factor (previously, only loocv was available)

RCC
-the function estim.regul has been renamed tune.rcc
-the function pcatune has been renamed tune.pca

Bug fixes:
----------
-in plotIndiv: horizontal and vertical abline set as a default argument
-a new argument in splsda() function added: near.zero.var = TRUE or FALSE to speed up computations (near.zero.var = FALSE to gain speed)
-the valid() function has been updated to speed up the computations. There is no 'criterion' argument to choose anymore (by default, all are included in the computation)
-in plotVar: matching arguments user-function to avoid additions of unused arguments
-in plotIndiv, arguments 'x.label' and 'y.label' were replaced by 'X.label' and 'Y.label'
-in pca, argument 'scale.' was changed to 'scale'

Changes in 4.1
================

New features:
-------------
- New S3 method valid for objects of class psl, spls, plsda and splsda
- New select.var function to directly extract the selected variables from spls, spca, sipca
- New data set vac18 for multilevel data


Changes in 4.0
================

New features:
-------------
- The multilevel methodology has been added as well as the associated S3 methods for the graphical outputs (plotVar, plotIndiv)
- pheatmap clustering is available for multilevel analysis (borrowed from the pheatmap package)
- tuning functions are available for multilevel analyses
- a dependency to the package 'igraph0' has been created (instead of 'igraph' as the authors informed us of major changes in this package)

Bug fixes:
----------
-pls and spls have been modified to better handle NA values


Changes in 3.0
================

New features:
-------------
- The new methodology IPCA and sIPCA have been added as well as the associated S3 methods for the graphical outputs

- GeneBank IDs and gene titles were added in the liver toxicity study


Changes in 2.9-6
================

New features:
-------------
- Modifying the valid function: the Q2 criterion has been implemented

- var.label argument is used in plotVar.plsda, plotVar.splsda, plot3dVar.plsda, plot3dVar.splsda instead of X.label

- New S3 method network for pls

- New code for valid function to PLS-DA and sPLS-DA models validation

- New code for plot.valid to display the results of the valid function for PLS-DA 
  and sPLS-DA models

- cim and network were modified to obtain the simMat matrix as value

- plotVar was modified to obtain the coordinates for X and Y variables as value

- In predict function, several or all prediction methods are available simultaneously to 
  predict the classes of test data with plsda and splsda 
  
- The argument 'mode' has been removed of plsda and splsda functions


Changes in 2.9-5
================

New features:
-------------
- sPCA has been modified to get orthogonal principal components


Changes in 2.9-4
================

New features:
-------------
- PCA has been modified to run either SVD (no missing values) or NIPALS (missing values)

- print.pca has been added to display the results of PCA

- pcatune has been added to guide the choice of the number of principal components


Changes in 2.9-1
================

New features:
-------------
- New S3 methods plotIndiv and plotVar for PCA

- New S3 method plot.valid to display the results of the valid function

- New code for imgCor function for a nicer representation of the correlation matrix

- In predict.plsda and predict.splsda functions the argument 'method' were replaced by
  method = c("max.dist", "class.dist", "centroids.dist", "mahalanobis.dist")

- New arguments for the cim function:
  * dendrogram
  * ColSideColors, RowSideColors

- Modifying the valid function:
  * missing data are allowed
  * Q2 criterion has been removed

- Functions pls, plsda, spls and splsda were modified to identify zero- or near-zero variance predictors

- Functions plotVar.plsda, plotVar.splsda, plot3dVar.plsda, plot3dVar.splsda were modified to represent 
  only the X variables

- New function: 'nearZeroVar' for identification of zero- or near-zero variance predictors 


Changes in 2.8-1
================

New features:
-------------
- New arguments ("axis.labelX", "axis.labelY") in the function imgCor, to indicate if the labels 
  of axis have to be shown or not

- New classes splsda and plsda for predict, print, plotIndiv, plot3dIndiv, plotVar, plot3dVar

- Several prediction functions are avaiable to predict the classes of test data with plsda and 
  splsda see predict (argument 'method' ("class.dist", "centroids.dist", "Sr.dist", "max.dist"))

- New functions map & unmap borrowed from the mclust package


Bug fixes:
----------


Changes in 2.7-1
================

New features:
-------------

- New functions pca, plsda and splsda, as well as extensions of plot3dVar and plot3dIndiv for pca 

- New network.default function which is called by network.rcc and network.spls

- bin.color function added in network.default to color edges w.r.t. the values in the simMat matrix

- nipals has been improved to be computationally more efficient

- Missing values are treated as in Tenenhaus in pls, spls and valid functions

- New argument 'ncomp' in rcc function, argument 'ncomp' has been removed from 'summary' and 'rcc'

- New option ("XY-variate") for the argument 'rep.space' in the 'plot3dVar'


Bug fixes:
----------
- 'tick marks' values have been corrected for color key in cim

- Computation of the simMat matrix for pls and spls - canonical mode, and correction in 
  plotVar, plot3dVar, cim and network

- Correction of the default argument 'rep.space = "XY-variate"' in plotIndiv and plot3dIndiv

- Correction of the manual


Changes in 2.6-0
================

New features:
-------------
- Former R package integrOmics has been renamed mixOmics

- In functions plotIndiv, plotVar, cim, network the arguments 'dim1', 'dim2', 'ncomp' 
  were replaced by 'comp', a vector of length 2 (by default 'comp = 1:2')

- Network has a new argument 'alpha'

User-visible changes:
---------------------

Bug fixes:
----------

Internal changes:
-----------------



