Changelog
Source:NEWS.md
fdacluster (development version)
- Expanded arguments of
fdakmeans()
to allow for more control over the type of input functional data:-
is_domain_interval
allows one to state the all curves are defined on the same fixed interval; -
transformation
specifies the transformation to be applied to the data before clustering. -
check_check_option_compatibility()
handles errors when incompatible options are selected.
-
- Created two separate C++ classes for L2 distance and normalized L2 distance; the former cannot be used in combination with dilation or affine warping classes because it is not invariant to these transformations.
- Integrated distances in C++ classes are now computed via
arma::trapz()
. - Added talk given at Rencontres R 2023 in Avignon, France to the News section of the website.
- Now uses package distops for computing distance matrices.
- Reduced number of dependencies: removed dplyr, tidyr, forcats, (progressr, tidyr, furrr, purrr).
- Updated
README
file. - Updated GHA workflows.
- Updated vignettes.
- Bug fixes.
fdacluster 0.3.0
CRAN release: 2023-07-04
- Added median centroid type;
- Median and mean centroid types are now defined on the union of individual grids;
- Simplified
caps
class to avoid storing objects multiple times under different names; - Added vignette on initialization strategies for k-means;
- Added article on use case about the Berkeley growth study;
- Added article on supported input formats.
fdacluster 0.2.2
CRAN release: 2023-05-25
- Make sure one can use fdacluster with namespace notation.
- Make sure not to use fda or funData before checking it is available.
fdacluster 0.2.1
CRAN release: 2023-05-19
- Add DBSCAN clustering;
- Fix C++ compiler issues that errored when accessing empty vectors.
fdacluster 0.2.0
CRAN release: 2023-05-18
- Add hierarchical clustering;
- Enforce
n_clusters
in output via linear programming (LP) using the lpSolve package; - New
caps
class for storing results from functional Clustering with Amplitude and Phase Separation in a consistent way; - Add tools for comparing clustering results (
mcaps
objects,autoplot
andplot
specialized method implementations); - Add seeding strategies for kmeans (via hierarchical clustering or k-means++ or k-means++ with exhaustive search of the first center or exhaustive search of all the centers);
- Add within-cluster domain auto-extension via mean imputation;
- Add possibility to cluster according to phase variability instead of amplitude variability.
- Renaming of functions: to perform k-means with alignment, now use
fdakmeans()
, to perform HAC with alignment, now usefdahclust()
.