mstknnclust: An R package for graph-based clustering via MST-kNN intersection

Unsupervised structure discovery text remains a central challenge in data mining, as most clustering algorithms require specifying the number of groups in advance or rely on density parameters difficult to tune. This paper presents mstknnclust , an R package implementing the MST- k NN algorithm, which partitions data by intersecting the Minimum Spanning Tree and the k -Nearest Neighbour graph. The value of k is determined automatically, requiring no user-specified hyperparameters. Benchmarking against well-known algorithms yields an Adjusted Rand Index of 0.65 without manual tuning, outperforming all parameter-free methods. The package is validated across computational linguistics, genomics, and seismology, and is freely available on CRAN.