https://github.com/cran/CluMix
Raw File
Tip revision: b6ad4d474d5865090b0bfabf2570aa79580237a4 authored by Manuela Hummel on 16 May 2017, 09:19:23 UTC
version 2.0
Tip revision: b6ad4d4
NEWS
CHANGES IN VERSION 2.3
----------------------

o Removed vignette due to problems with encoding. Please refer to vignette in version 2.2


CHANGES IN VERSION 2.2
----------------------

o Bug fix in 'dist.subjects' and 'dist.variables' regarding presence of character variables in the data 

o Added CITATION


CHANGES IN VERSION 2.1
----------------------

o Bug fix in 'mix.heatmap' (weird behaviour when there was exactly 1 binary variable)

o In 'distmap' main diagonal points now are shown with dark color (similarity=1) for both samples and variables (for samples before they were white)

o Updated links to functions from 'ClustOfVar' package in the documentation

o Vertical legend matrix labels are now possible in 'mix.heatmap' via the new argument 'legend.srt'


CHANGES IN VERSION 2.0
----------------------

o A new additional method for calculating distances between variables based on distance correlation was introduced in 'mix.heatmap', 'dist.variables', 'similarity.variables', 'dendro.variables', 'distmap', and 'confounderPlot'. 

o The agglomeration method for hierarchical clustering in 'mix.heatmap', 'dendro.subjects', 'dendro.variables' and 'distmap' can now be chosen via the argument 'linkage' (equivalent to argument 'method' in 'hclust'). The default is now 'ward.D2' and not 'complete' anymore.  

o Parameter 'alwaysGower' in 'dist.subjects' now allows to calculate Gower's distances even in the case of exclusively continuous data. 


CHANGES IN VERSION 1.3
----------------------

o With the new argument 'cont.fixed.range' in 'mix.heatmap' the user can decide to apply for all quantitative variables the same color range, i.e. variables with a smaller range compared to the overall range will not exhibit all colors given by 'col.cont'. Additionally, the maximal values corresponding to "extreme colors" can be manipulated by the new argument 'cont.range'.
The previous and now default ('cont.fixed.range = FALSE') behaviour is that every quantitative variable is visualized independently using all colors of 'col.cont' (assuming variables of different origins with potentially completely different distributions). Argument 'cont.range' is ignored in this case.  

o The legend for quantitative variables in 'mix.heatmap' now shows a color gradient. If 'cont.fixed.range = TRUE', the actual extreme values are shown, either as calculated from the data or as given by 'cont.range'.  

o The package vignette is now written with knitr.
 

CHANGES IN VERSION 1.2
----------------------

o Fix bug in column/row bar legend (order of colors)  


CHANGES IN VERSION 1.1
----------------------

o Added options 'weights' or 'varweights' to 'dist.subjects', 'similarity.subjects', 'dendro.subjects', 'distmap' and 'mix.heatmap' to allow for variable weights in the calculation of Gower's distances between subjects


CHANGES IN VERSION 1.0
----------------------

o Added package vignette

o Added function 'similarity.subjects'


CHANGES IN VERSION 0.5
----------------------

o Changed default coloring

o Binary variables that are logical or numeric are now recognized as binary


CHANGES IN VERSION 0.4
----------------------

o Reordering the dendrogram for 'distmap' is now possible via the new option 'reorderdend'. It works in the same way as the 'wts' option in the reorder.dendrogram function from the stats package

o The 'confounderPlot' is now more flexible w.r.t. xlim, ylim, plotting color, symbol and font.
back to top