Rclusterpp provides flexible native hierarchical clustering routings optimized for performance and minimal memory requirements. In particular Rclusterpp includes "stored data" clustering implementations with O(n) memory footprints. Rclusterpp has been successfully used to cluster 100,000s of observations.
Rclusterpp makes extensive use of
Rcpp for integration with R, and
the Eigen matrix library (via
RcppEigen).
Rclusterpp provides a R interface to its internal libraries that can be used in
place of stats::hclust and provides linkable libraries for use by downstream packages.
Explore the unit tests inst/unit_tests and examples directory inst/examples
for examples on how to use Rclusterpp directly within R, or as a linkable library
for use with other native code. Note that some of the examples
require the inline package.
Rclusterpp uses OpenMP internally for concurrent execution. By default, as many
threads as processors are created. To control the number of threads set the
OMP_NUM_THREADS environment variable.
Rclusterpp installation instructions can be found on the project wiki.