mstknnclust 1.0.0
- First major release.
- kNN construction uses a partial-sort approach that stores only
floor(log(n)) neighbours per node, yielding O(n^2 log n)
time and O(n * log(n)) working memory.
- MST is computed via
igraph::mst() on a sparse edge list
constructed from integer index vectors.
- All deprecated igraph function calls replaced with modern
equivalents (
components(), intersection(),
is_connected(), etc.).
- Package structure reorganised following R Packages (2e) conventions:
one file per exported function, testthat test suite, pkgdown site.
- Computational performance benchmarks added (see
inst/extdata/benchmark_performance.R).
- Added pkgdown website at https://jorgeklz.github.io/package-mstknnclust/.
mstknnclust 0.3.2