# Christian Hennig

#### 8 packages on CRAN

Various methods for clustering and cluster validation. Fixed point clustering. Linear regression clustering. Clustering by merging Gaussian mixture components. Symmetric and asymmetric discriminant projections for visualisation of the separation of groupings. Cluster validation statistics for distance based clustering including corrected Rand index. Standardisation of cluster validation statistics by random clusterings and comparison between many clustering methods and numbers of clusters based on this. Cluster-wise cluster stability assessment. Methods for estimation of the number of clusters: Calinski-Harabasz, Tibshirani and Walther's prediction strength, Fang and Wang's bootstrap stability. Gaussian/multinomial mixture fitting for mixed continuous/categorical variables. Variable-wise statistics for cluster interpretation. DBSCAN clustering. Interface functions for many clustering methods implemented in R, including estimating the number of clusters with kmeans, pam and clara. Modality diagnosis for Gaussian mixtures. For an overview see package?fpc.

Distance-based parametric bootstrap tests for clustering with spatial neighborhood information. Some distance measures, Clustering of presence-absence, abundance and multilocus genetic data for species delimitation, nearest neighbor based noise detection. Genetic distances between communities. Tests whether various distance-based regressions are equal. Try package?prabclus for on overview.

Some M-estimators for 1-dimensional location (Bisquare, ML for the Cauchy distribution, and the estimators from application of the smoothing principle introduced in Hampel, Hennig and Ronchetti (2011) to the above, the Huber M-estimator, and the median, main function is smoothm), and Pitman estimator.

Offers a set of functions for extending 'dendrogram' objects in R, letting you visualize and compare trees of 'hierarchical clusterings'. You can (1) Adjust a tree's graphical parameters - the color, size, type, etc of its branches, nodes and labels. (2) Visually and statistically compare different 'dendrograms' to one another.

Performs robust cluster analysis allowing for outliers and noise that cannot be fitted by any cluster. The data are modelled by a mixture of Gaussian distributions and a noise component, which is an improper uniform distribution covering the whole Euclidean space. Parameters are estimated by (pseudo) maximum likelihood. This is fitted by a EM-type algorithm. See Coretto and Hennig (2016) <doi:10.1080/01621459.2015.1100996>, and Coretto and Hennig (2017) <http://jmlr.org/papers/v18/16-382.html>.

Various quantile-based clustering algorithms: algorithm CU (Common theta and Unscaled variables), algorithm CS (Common theta and Scaled variables through lambda_j), algorithm VU (Variable-wise theta_j and Unscaled variables) and algorithm VW (Variable-wise theta_j and Scaled variables through lambda_j). Hennig, Viroli, Anderlucci (2018) <arXiv:1806.10403v1>.

Trimmed k-means clustering. The method is described in Cuesta-Albertos et al. (1997) <doi:10.1214/aos/1031833664>.