NEWS
Version 0.1.0.1
2024-05-10 1. Update the test cases to be consistent with testthat (>= 3.0.0).
2024-04-17 1. Created version 0.1.0.1 from 0.1.0. 1. Fixed a testthat testing issue caused by a new upcoming version of the ‘dqrng’ package, as notified by the author of the package.
Version 0.1.0
2022-01-25
- When cluster_method = “Ball+BIC”, if Mclust discretization was unsuccessful, discretize.jointly will now add jitter noise (of increasing ‘amount’) to ‘data’.
2022-01-18
- Updated CITATION, REFERENDCES.bib, and README.md.
2022-01-17
- Fixed a bug in computing weights for MultiChannel.WUC when cluster IDs are not consecutive numbers after Ball clustering by Mclust (descrtize_jointly.R:198-204)
- Fixed function cluster(data, k, method) where the method was not included into the return object.
- Examples in the vignettes have been updated.
2022-01-11
- Added k=Inf option, that incrementally increases the number of clusters to choose until either an increment is not required or each data point is its own cluster.
- Added an example showing the use of k=Inf option.
2021-10-26
- Added an additional parameter ‘cluster_method’, could be “Ball+BIC”, “kmeans+silhouette” or “PAM”,
- Added clustering method Mclust
- Changed the default cluster method from Kmeans to Mclust, which is “Ball+BIC”
- Added an additional parameter ‘grid_method’, could be “Sort+split” or “MultiChannel.WUC”, “MultiChannel.WUC” is a discretize method Multichannel.MUC from Ckmeans.1d.dp. The default option is still “Sort+split” which is the same as previous version. As the “MultiChannel.WUC” option is experimental, it is not recommended.
- Added general plotting function plot(), old plot function still available
- Changed the default value of min_level from 2 to 1. We expanded min_level argument to provide a vector of integers specifying the minimum level for each dimension.
Version 0.0.8.1
2021-10-06
- Removed the requirement for the number of points in the data
Version 0.0.8
2020-09-13
- Added function plotGOCpatterns to plot the continuous data along with the cluster preserving grid.
- Created a manual for the plotGOCpatterns() function.
- Updated the code for the Examples vignette to use plotGOCpatterns.
2020-08-10
- Created version 0.0.8 from 0.0.7.
- Added an additional parameter ‘min_level’ to denote the minimum number of discretization levels required for each dimension.
- Updated the manual of discrete.jointly() function.
- Added an entry in reference and citation.
- Updated README.md with badges.
Version 0.0.7
2020-04-03
- Tidied up code for the Examples vignette.
- Updated the manual of discrete.jointly() function.
- Made minor editorial changes in DESCRIPTION and README.md.
- Resolved signed/unsigned mismatches.
2020-03-31
- Created version 0.0.7 from 0.0.6.
- Fixed memory leak in Clusters.cpp when calculating median.
Version 0.0.6
2020-03-26
- Created version 0.0.6 from 0.0.5.
- Rewrote Prep_Index() to work in between two consecutive points, rather than on top of a single point.
- Using distance() in Prep_Index() to calculate the distance for two iterators.
- Using “ceil” in Binary_Index_Searching() to consider even/odd cases when determining grid lines.
- Fixed potential overflow issues.
Version 0.0.5 (not released to the public)
2020-03-25
- Fixed a bug in the prep_index() function.
- Fixed prep_index() (lines 120 and 125) such that grid lines are put at the midpoint between two conseuctive points, instead of on one of the points.
- Updated vignette. Example 2 seems always correct now.
2020-03-24
- Created version 0.0.5 from 0.0.4.
- Function discretize.jointly() now returns cluster labels of each observation and a similarity score (ARI) between the joint discretization and the cluster labels of each observation.
- The class Cluster has a new constructor that takes cluster labels and the input data to compute median for each dimension.
- Find_grid() is now based on median.
- Using ‘dqrng’ in test cases to avoid RNG issue in testing.
- Rewrote multiple functions in Joint_Grid.cpp to avoid push_back().
- New visualization code in vignette now shows cluster labels for each observation.
Version 0.0.4 (not released to the public)
2020-03-20
- Created version 0.0.4 from 0.0.3.
- Fixed typos in DESCRIPTION and README files.
Version 0.0.3
2020-03-17
- Created version 0.0.3 from 0.0.2. Package renamed to GridOnClusters
- Function joint.grid.discretize.R() renamed to discretize.jointly()
- Return values of function discretize.jointly() changed to include both the discretized data and the grid
- Manual for discretize.jointly() updated.
- Line 104, 105 in Joint_Grid.cpp commented out
- Rewrote Find_Grid() to avoid push_back() in Joint_Grid.cpp
- Created a vignette to include examples.
Version 0.0.2 (not released to the public)
2020-03-14
- Created the initial version 0.0.1. Package renamed to QNJGD
Version 0.0.1 (not released to the public)
2020-03-09
- Created the initial version 0.0.1. Package named JointGridDiscr