The R Journal: Clustering Binary Data Optimizing the Exclusiveness of Features
- Citation:
- Cerdeira, Orestes J., and Tiago Monteiro-Henriques. "The R Journal: Clustering Binary Data Optimizing the Exclusiveness of Features." The R Journal. 18 (2026): 329-343.
Abstract:
A problem that has been studied for a long time in Vegetation Science consists in finding distinct groups among a collection of vegetation samples (relevés). Recently a criterion has been proposed to capture the patterns of differential species among the groups from any arbitrary M-cluster of relevés. The criterion optimization is quite complex (NP-hard) as it implies searching a huge set of possible combinations of relevés into M groups. In this paper we give an integer linear programming formulation to optimize the criterion for M=2, that can be used to solve moderately-sized data problems; we also describe a greedy randomized adaptive search procedure and a simulated annealing algorithm for arbitrary M. We combined these different approaches and prepared a collection of software functions, which are available in the open-source R package diffval. We illustrate the use of such functions using real-world data.
Notes:
