Clusters state sequences and weighted data. It provides an optimized weighted PAM algorithm as well as functions for aggregating replicated cases, computing cluster quality measures for a range of clustering solutions and plotting (fuzzy) clusters of state sequences. Parametric bootstraps methods to validate typology of sequences are also provided.
CHANGES IN WeightedCluster VERSION 1.4
NEW FEATURES o seqpropclust(), wcPropertyClustering(): Monothetic clustering with the DIVCLUS-T algorithm. Automatic sequence properties extraction. o as.clustrange(): Handle monothetic clustering.
CHANGES IN WeightedCluster VERSION 1.2-1
SMALL FIXES o Fixed import related notes in RCMD check.
MISC: o Registration of native routines for R 3.4.x and TraMineR >= 2.0-6. See https://stat.ethz.ch/pipermail/r-devel/2017-February/073755.html.
CHANGES IN WeightedCluster VERSION 1.2
NEW FEATURES o as.clustrange(), wcKMedRange(): Added bootstrap of cluster quality measures. o wcCmpCluster(): Automatically compute different clustering solutions and associated quality measures to help identifying the best one. o as.clustrange(), wcKMedRange(): Speed and memory handling improvements.
BUG FIXES o wcKMedRange() was internally converting dist objects to matrix. For this reason, it used too much memory. This isn't the case anymore. o Moved the vignettes source to the vignettes folder.
CHANGES IN WeightedCluster VERSION 1.0
NEW FEATURES o The vignette (English version) has been published in the LIVES Working papers series. o The vignette (manual) has been translated in English. o seqclustname: automatic labeling of clusters using sequence medoids. o Added the WeightedCluster vignette (in French). o It is now possible to plot the "RHC" statistic computed as 1-HC. This way all statistics should be maximized (Requested by Gilbert Ritschard). o It is now possible to specify the colors used by plot.clustrange (Requested by Sandra Ham). o Now testing for correct input in "as.clustrange".
BUG FIXES o PBC, HG, HC, where computed only on the lower triangular part of the distance matrix (instead on the whole one). o Corrected citation informations o Many grammar and spelling corrections in the WeightedClusterPreview vignette (thanks to Nevena Zhelyazkova).
CHANGES IN WeightedCluster VERSION 0.9
o First development release