A comparative study of hard clustering algorithms for vegetation data

Abstract

Questions Which clustering algorithms are most effective according to different cluster validity evaluators? Which distance or dissimilarity measure is most suitable for clustering algorithms? Location Hyrcanian forest, Iran (Asia), Virginia region forest, United States (North America), beech forests, Ukraine (Europe). Methods We tested 25 clustering algorithms with nine vegetation data sets comprised of three real data sets and six simulated data sets exhibiting different cluster separation values. The clustering algorithms included both hierarchical and non‐hierarchical partitioning. Five evaluators were employed on each cluster solution to evaluate different clustering algorithms. Algorithms were ranked from best to worst on each clustering evaluator for each data set. Results The comparison revealed that the OPTSIL initiated from a Flexible‐β (−0.25) solution achieved particularly good performance. We also found that Ward's method and Flexible‐β (−0.1) implementations were accurate. K‐means with Hellinger distance was superior to Partitioning Around Medoids (PAM) algorithms. Accordance between distance measures and clustering algorithms was also observed. Bray–Curtis dissimilarity combined with a range of clustering algorithms was successful in most cases. Bray–Curtis dissimilarity proved superior to other distance measures for heterogeneous data sets. Conclusions All in all, the results demonstrate that choosing the most suitable method before clustering is critical for achieving maximally interpretable clusters. The complexity of vegetation data sets makes this issue even more important. The choice of distance measure had more effect than the choice of clustering method on the quality of results. Our results illustrate that OPTSIL Flexible‐β (−0.1) and OPTPART could prove superior to alternative conventional clustering algorithms when internal evaluation criteria are used to optimize clustering.

Can vegetation be discretely classified in species‐poor environments? Testing plant community concepts for vegetation monitoring on sub‐Antarctic Marion Island

“…The classifications were undertaken in three steps: (1) pre‐processing involved the selection of a distance measure and normalization of the data; (2) cluster analysis involved the selection and application of the clustering algorithm and its various parameters; (3) cluster validation involved the selection and application of appropriate internal validation techniques to evaluate the quality of the classification. Four clustering algorithms and four validation measures were explored based on demonstrated performance in recent literature (Aho et al, 2008 ; Handl et al, 2005 ; Lengyel et al, 2021 ; Pakgohar et al, 2021 ). We defined a vegetation classification as being comprised of a cluster of plots organized into units with discrete boundaries between them.…”

“…Lötter et al ( 2013 ) referred to this as “the classification conundrum”. The amount of research available which advocates particular methods, ideologies and approaches to classify vegetation (Feilhauer et al, 2020 ; Lengyel et al, 2021 ; Lortie et al, 2004 ; Lötter et al, 2013 ; Pakgohar et al, 2021 ), reflects the impracticality of the use of one universal approach in all environments. Nevertheless, there is general agreement that expert opinion is needed to select vegetation units at some stage in the classification process (Brown et al, 2013 ; Lötter et al, 2013 ; Mucina, 1997 ) even if this adds subjectivity to the classification, possibly resulting in bias (Lötter et al, 2013 ; Wolda, 1981 ), with little objective validation of clustering results.…”

“…Nevertheless, there is general agreement that expert opinion is needed to select vegetation units at some stage in the classification process (Brown et al, 2013 ; Lötter et al, 2013 ; Mucina, 1997 ) even if this adds subjectivity to the classification, possibly resulting in bias (Lötter et al, 2013 ; Wolda, 1981 ), with little objective validation of clustering results. However, recent classification methods, especially those used in data science (Flynt & Dean, 2016 ), have made it possible to formally test the effectiveness of classifications, thereby reducing the number of subjective choices (Lötter et al, 2013 ; Pakgohar et al, 2021 ). The existence of discrete groups in the data can thus be tested objectively, before expert interpretation is needed.…”

Boreal moss-microbe interactions are revealed through metagenome assembly of novel bacterial species

“…To obtain community assembly matrices, for VST transformed data, we used Jaccard dissimilarity index calculated from presence-absence data (Tian et al, 2019); while for Hellinger transformed data, we used the Euclidean dissimilarity index (Pakgohar et al, 2021).…”

Individual tree-based forest species diversity estimation by classification and clustering methods using UAV data

“…Regression analysis is to model the spectral and structural information directly with the measured species diversity indices, which is a mature and straightforward algorithm, but the applicability in different regions is poor (Ceballos et al, 2015). The clustering algorithm can evaluate species diversity by grouping trees with similar characteristics based on the biochemical and structural variation of different tree species (Asner et al, 2015;Padilla-Martinez et al, 2020;Pakgohar et al, 2021). Clustering can be used to identify patterns or trends in the distribution and abundance of different species within a forest ecosystem.…”