Changelog
Source:NEWS.md
rainette (development version)
- Add
colors
argument torainette_plot()
to allow to specify custom colors for clusters (#37, thanks @lvaudor)
rainette 0.3.1
CRAN release: 2023-03-17
- Replace call to
palette.colors()
with manual colors vector to ensure compatibility with R 3.6 (#11) - Fix dendrogram not displayed when show_na_title = FALSE
rainette 0.3.0
CRAN release: 2022-02-18
New features
Rework of the double classification computation :
- New
full
argument torainette2()
. IfTRUE
(default, same behavior as previous versions), all crossings between groups of the two single classifications are taken into account to find the best partition. IfFALSE
, only the crossings with maximal associations are kept. - New
parallel
argument torainette2()
to compute partitions withmclapply
(FALSE
by default, won’t work on Windows, uses more RAM) - Global optimization and speed up of
rainette2()
computations, with added progress bars to better estimate long runs
Documentation
- Improved french vignette “description des algorithmes”
- New english vignette “algorithms description”
- Reworked french and english introduction vignettes
Deprecated features
- The wordcloud plots will be deprecated in a near future. A warning has been added to
rainette_plot()
andrainette2_plot()
if they are called withtype = "cloud"
.
Other
- Add
show_na_title
,cluster_label
andkeyness_plot_xlab
arguments torainette_plot()
to customize graphics output - Fix warnings in Font Awesome icon names
rainette 0.2.1
CRAN release: 2021-10-06
- Add option to show merged segments in document browser
- Fix warning and error in
rainette_explor
andrainette2_explor
when a cluster dfm is empty - Fix error when the dfm contains empty string as feature
rainette 0.2.0
CRAN release: 2021-06-25
Important and breaking changes
-
min_uc_size
,uc_size1
anduc_size2
arguments torainette
andrainette2
have been renamed tomin_segment_size
,min_segment_size1
andmin_segment_size2
. - The default value of
min_segment_size
inrainette
is now0
, which means that no merging is done between segments by default. Results could then be different from previous package versions whenmin_uc_size
was not specified. - Important bugfix : merging of segments based on
min_segment_size
was not handled correctly in the previous versions regarding the segment sources, as segments from different documents could be merged together. This should now be fixed.
New features
- A new graphical interface to browse cluster documents has been added to
rainette_explor
andrainette2_explor
. - New function
clusters_by_doc_table
which gives the number of segments of each cluster for each document. - New function
docs_by_cluster_table
which gives, for each cluster, the number of documents with at least one segment in this cluster. -
split_segments
should now be about 4 times faster. - Terms frequencies and documents proportions statistics have been added to the explor interfaces.
Other
- When
rainette
is called withmin_segment_size
> 0, adoc_id
argument must be given which is the name of adtm
docvar identifying the segments source. If the corpus has been produced bysplit_segments
, the addedsegment_source
docvar is used by default. - Color palette for clusters changed to “Tableau 10”.
- Negative keyness values are not shown by default anymore in
rainette_explor
,rainette2_explor
,rainette_plot
andrainette2_plot
. - Wordcloud plots have been removed from explor interfaces.
- A warning is displayed when
min_split_members
< 3. - If
rainette_explor
is called on arainette2
results object,rainette2_explor
is launched automatically.
rainette 0.1.3
CRAN release: 2021-05-11
- Parallel computing in split_segments has been removed
- Fix potential name conflicts in
rainette_explor
- Compatibility with quanteda v3 (thanks @kbenoit)