Prediction and Clustering on the Torus by Conformal Prediction
Pairwise L2 angular distance
Clustering on the torus by conformal prediction
Angular distance
Angular subtraction
Clustering by connected components of ellipsoids
ClusTorus: Prediction and Clustering on the Torus by Conformal Predict...
Conformal prediction set indices with kernel density estimation
K-Means Clustering to K-Spheres Clustering on Torus
Fitting mixtures of bivariate von Mises distribution
Grid on torus
Selecting optimal level based on the runs of the number of clusters
Selecting optimal number of mixture components based on various criter...
Selecting optimal hyperparameters for the conformal prediction set
Inductive prediction sets for each level
Conformity score for inductive prediction sets
Kernel density estimation using circular von Mises distribution
K-Means Clustering on Torus
Transform the angular data to be on principal interval
Toroidal subtraction
Weighted extrinsic mean direction and mean resultant length
Provides various tools of for clustering multivariate angular data on the torus. The package provides angular adaptations of usual clustering methods such as the k-means clustering, pairwise angular distances, which can be used as an input for distance-based clustering algorithms, and implements clustering based on the conformal prediction framework. Options for the conformal scores include scores based on a kernel density estimate, multivariate von Mises mixtures, and naive k-means clusters. Moreover, the package provides some basic data handling tools for angular data.
Useful links