pith. sign in

arxiv: 2409.13760 · v1 · pith:YGTKGT7Bnew · submitted 2024-09-16 · 📊 stat.AP

Mapping climate change awareness through spatial hierarchical clustering

Gianpaolo Zammarchi , Paolo Maranzano This is my paper

Pith reviewed 2026-05-23 21:02 UTC · model grok-4.3

classification 📊 stat.AP
keywords climate change awarenesshierarchical clusteringspatial clusteringcountry groupingWard algorithmgeographical distancessocio-economic indicators
0
0 comments X

The pith

Adding physical distances between countries to clustering yields more stable and compact groups by climate awareness.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper applies a Ward-like hierarchical clustering method to countries using data on climate change awareness, socio-economic factors, climate characteristics, and physical distances. A custom tuning procedure selects hyperparameters by evaluating within-cluster homogeneity, between-cluster separation, and directly comparing results from versions that include versus exclude the geographical distances. The geographically-informed version produces partitions with greater stability and yields aggregations that are both more interpretable and more compact in space. The resulting groups show Western countries forming a high-awareness cluster while Asian, African, and Middle Eastern countries display lower average awareness with greater internal variability.

Core claim

By combining climate change awareness measures, socio-economic factors, climate-related characteristics, and physical distances in a Ward-like hierarchical clustering algorithm, and using a customized hyperparameter selection that compares geographical and non-geographical partitions, the analysis identifies more stable, interpretable, and geographically-compact country groups, highlighting high awareness in Western countries versus lower awareness elsewhere.

What carries the argument

Geographically-informed Ward-like hierarchical clustering algorithm that incorporates physical distances alongside awareness and socio-economic features.

If this is right

  • The geographically-informed partitions exhibit greater stability than those based solely on awareness and socio-economic data.
  • Western countries form one high-awareness and geographically compact cluster.
  • Asian, African, and Middle Eastern countries form groups with greater variability but overall lower awareness.
  • The spatial component produces aggregations that are easier to interpret because they respect physical proximity.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The identified clusters could support region-specific strategies for raising climate awareness.
  • The same spatial clustering approach might be tested on other cross-country attitude datasets such as public health or economic policy views.
  • Re-running the analysis on newer awareness surveys would show whether the Western versus other-regions contrast persists over time.

Load-bearing premise

The climate change awareness measures, socio-economic indicators, and climate characteristics are accurate, comparable across countries, and relevant without systematic measurement error or selection bias.

What would settle it

Running the same data through a non-geographical clustering and finding equal or greater stability in the partitions, or obtaining awareness groups that shift markedly when independent survey data replace the current measures.

Figures

Figures reproduced from arXiv: 2409.13760 by Gianpaolo Zammarchi, Paolo Maranzano.

Figure 1
Figure 1. Figure 1: World map representing the share of people in the 2022 survey with medium-high and high climate change [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Comparison between awareness measured in 2021 and 2022 [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Optimal values of the mixing parameter α conditioning on K clusters (that is, α ∗ K) obtained using the criterion by [27] (α ∗ K,max in orange) and by [20] (α ∗ K,min in blue). [27]. Below we present the results related to these two cases when considering the geographical constraint and when ignoring the spatial dimension. Figures 6 through 9 show two sets of maps produced for K = 4 and K = 5 with α = 0 or… view at source ↗
Figure 4
Figure 4. Figure 4: Estimates of the Silhouette index, the Dunn’s index, the C-index, the Calinski-Harabasz’s index, and the [PITH_FULL_IMAGE:figures/full_fig_p010_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Estimated percentage gain (positive values) or loss (negative values) of the Silhouette, Dunn’s, C-index, [PITH_FULL_IMAGE:figures/full_fig_p011_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Map of clustering partitions obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p012_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Map of clustering partitions obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p012_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Map of clustering partitions obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p013_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Map of clustering partitions obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p013_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Descriptive statistics on the share of interviewed people declaring a low or medium-low climate change [PITH_FULL_IMAGE:figures/full_fig_p014_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Heatmap of pairwise Pearson’s linear correlation index for the full set of climate awareness, climate-related [PITH_FULL_IMAGE:figures/full_fig_p015_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Heatmap of pairwise Adjusted Rand Index for the four clusterings from the main analysis (marked as [PITH_FULL_IMAGE:figures/full_fig_p016_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: Optimal values of the mixing parameter α conditioning on K clusters (that is, α ∗ K) obtained using the criterion by Morelli et al. (2024) (α ∗ K,max in orange) and by Chavent et al. (2018) (α ∗ K,min in blue). 19 [PITH_FULL_IMAGE:figures/full_fig_p019_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: Estimates of the Silhouette index, the Dunn’s index, the C-index, the Calinski-Harabasz’s index, and the [PITH_FULL_IMAGE:figures/full_fig_p020_14.png] view at source ↗
Figure 15
Figure 15. Figure 15: Estimated percentage gain (positive values) or loss (negative values) of the Silhouette, Dunn’s, C-index, [PITH_FULL_IMAGE:figures/full_fig_p021_15.png] view at source ↗
Figure 16
Figure 16. Figure 16: Map of clustering partitions obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p022_16.png] view at source ↗
Figure 17
Figure 17. Figure 17: Map of clustering partitions obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p022_17.png] view at source ↗
Figure 18
Figure 18. Figure 18: Map of clustering partitions obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p023_18.png] view at source ↗
Figure 19
Figure 19. Figure 19: Map of clustering partitions obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p023_19.png] view at source ↗
Figure 20
Figure 20. Figure 20: Descriptive statistics on the share of interviewed people declaring a low or medium-low climate change [PITH_FULL_IMAGE:figures/full_fig_p024_20.png] view at source ↗
Figure 21
Figure 21. Figure 21: Optimal values of the mixing parameter α conditioning on K clusters (that is, α ∗ K) obtained using the criterion by Morelli et al. (2024) (α ∗ K,max in orange) and by Chavent et al. (2018) (α ∗ K,min in blue). 25 [PITH_FULL_IMAGE:figures/full_fig_p025_21.png] view at source ↗
Figure 22
Figure 22. Figure 22: Estimates of the Silhouette index, the Dunn’s index, the C-index, the Calinski-Harabasz’s index, and the [PITH_FULL_IMAGE:figures/full_fig_p026_22.png] view at source ↗
Figure 23
Figure 23. Figure 23: Estimated percentage gain (positive values) or loss (negative values) of the Silhouette, Dunn’s, C-index, [PITH_FULL_IMAGE:figures/full_fig_p027_23.png] view at source ↗
Figure 24
Figure 24. Figure 24: Map of clustering partitions obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p028_24.png] view at source ↗
Figure 25
Figure 25. Figure 25: Map of clustering partitions obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p028_25.png] view at source ↗
Figure 26
Figure 26. Figure 26: Map of clustering partitions obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p029_26.png] view at source ↗
Figure 27
Figure 27. Figure 27: Map of clustering partitions obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p029_27.png] view at source ↗
Figure 28
Figure 28. Figure 28: Map of clustering partitions obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p030_28.png] view at source ↗
Figure 29
Figure 29. Figure 29: Map of clustering partitions obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p030_29.png] view at source ↗
Figure 30
Figure 30. Figure 30: Map of clustering partitions obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p031_30.png] view at source ↗
Figure 31
Figure 31. Figure 31: Map of clustering partitions obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p031_31.png] view at source ↗
Figure 32
Figure 32. Figure 32: Map of clustering partitions obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p032_32.png] view at source ↗
Figure 33
Figure 33. Figure 33: Map of clustering partitions obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p032_33.png] view at source ↗
Figure 34
Figure 34. Figure 34: Map of clustering partitions obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p033_34.png] view at source ↗
Figure 35
Figure 35. Figure 35: Map of clustering partitions obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p033_35.png] view at source ↗
Figure 36
Figure 36. Figure 36: Map of clustering partitions obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p034_36.png] view at source ↗
Figure 37
Figure 37. Figure 37: Map of clustering partitions obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p034_37.png] view at source ↗
Figure 38
Figure 38. Figure 38: Map of clustering partitions obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p035_38.png] view at source ↗
Figure 39
Figure 39. Figure 39: Map of clustering partitions obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p035_39.png] view at source ↗
read the original abstract

Climate change is a critical issue that will be in the political agenda for the next decades. While it is important for this topic to be discussed at higher levels, it is also of paramount importance that the populations became aware of the problem. As different countries may face more or less severe repercussions, it is also useful to understand the degree of awareness of specific populations. In this paper, we present a geographically-informed hierarchical clustering analysis aimed at identify groups of countries with a similar level of climate change awareness. We employ a Ward-like clustering algorithm that combines information pertaining climate change awareness, socio-economic factors, climate-related characteristics of different countries, and the physical distances between countries. To choose suitable values for the clustering hyperparameters, we propose a customized algorithm that takes into account the within-clusters homogeneity, the between-clusters separation and that explicitly compares the geographically-informed and non-geographical partitioning. The results show that the geographically-informed clustering provides more stability of the partitions and leads to interpretable and geographically-compact aggregations compared to a clustering in which the geographical component is absent. In particular, we identify a clear contrast among Western countries, characterized by high and compact awareness, and Asian, African, and Middle Eastern countries having greater variability but still lower awareness.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 2 minor

Summary. The paper claims that a geographically-informed hierarchical clustering (Ward-like algorithm incorporating climate awareness, socio-economic factors, climate characteristics, and physical distances) yields more stable partitions and geographically compact, interpretable country groups than a non-geographical version. A custom hyperparameter algorithm evaluates within-cluster homogeneity, between-cluster separation, and explicitly compares the two variants; results highlight a contrast between high-awareness Western countries and lower-awareness Asian/African/Middle Eastern ones.

Significance. If the stability and compactness advantages can be shown to hold under independent hyperparameter selection, the approach would offer a practical tool for identifying awareness patterns using external country-level data and a standard clustering objective. The explicit comparison of geo vs. non-geo variants is a methodological strength when properly validated, but the current lack of quantitative metrics reduces the immediate utility for policy or further research.

major comments (1)
  1. [Abstract and Methods] Abstract and Methods (description of customized hyperparameter algorithm): the procedure selects the number of clusters and distance weighting by explicitly comparing geo-informed and non-geographical partitions on homogeneity/separation criteria. This makes the reported gains in stability potentially circular, as the selection itself favors the geo variant; the manuscript must demonstrate the stability advantage with hyperparameters chosen independently (e.g., via a fixed criterion or separate validation set) rather than through the same comparison used to assert superiority.
minor comments (2)
  1. [Abstract] Abstract: no source, sample size, or quantitative stability metric (e.g., adjusted Rand index or silhouette score) is provided for the awareness data or the stability claim.
  2. [Results] Results: the final values chosen for the number of clusters and the geographical weighting parameter are not stated explicitly.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for highlighting a methodological concern regarding potential circularity in our hyperparameter selection. We agree this requires clarification and revision to strengthen the validity of our stability comparisons.

read point-by-point responses

  1. Referee: [Abstract and Methods] Abstract and Methods (description of customized hyperparameter algorithm): the procedure selects the number of clusters and distance weighting by explicitly comparing geo-informed and non-geographical partitions on homogeneity/separation criteria. This makes the reported gains in stability potentially circular, as the selection itself favors the geo variant; the manuscript must demonstrate the stability advantage with hyperparameters chosen independently (e.g., via a fixed criterion or separate validation set) rather than through the same comparison used to assert superiority.

    Authors: We acknowledge the validity of this concern: the customized algorithm's explicit comparison during selection could bias the reported stability advantages toward the geo-informed variant. In the revised manuscript, we will modify the Methods section to select the number of clusters (k) and geographic distance weighting using an independent, standard criterion (e.g., silhouette score or elbow method based on within-cluster homogeneity) applied separately to the geo-informed and non-geographical variants, without any cross-comparison in the selection step. Stability metrics (e.g., adjusted Rand index on bootstrap resamples) will then be computed and compared post-selection. The abstract, results, and discussion will be updated to describe this revised procedure and present the corresponding findings. This change ensures the stability evaluation is non-circular while preserving the core contribution of comparing the two clustering approaches. revision: yes

Circularity Check

1 steps flagged

Hyperparameter tuning explicitly compares geo vs non-geo partitions, making stability/compactness superiority claim reduce to the selection procedure

specific steps
  1. fitted input called prediction [Abstract]
    "To choose suitable values for the clustering hyperparameters, we propose a customized algorithm that takes into account the within-clusters homogeneity, the between-clusters separation and that explicitly compares the geographically-informed and non-geographical partitioning. The results show that the geographically-informed clustering provides more stability of the partitions and leads to interpretable and geographically-compact aggregations compared to a clustering in which the geographical component is absent."

    The hyperparameter selection procedure is defined to include an explicit geo vs non-geo comparison plus homogeneity/separation metrics; the paper then presents the outcome of that tuned comparison as evidence that geo-informed clustering is superior in stability and compactness. The claimed advantage is therefore generated by the same procedure used to select the model, rather than evaluated independently.

full rationale

The paper's central claim of geo-informed superiority rests on a customized hyperparameter algorithm whose selection criterion itself performs the geo vs non-geo comparison and homogeneity/separation evaluation. Because the tuning procedure incorporates the exact contrast later asserted as an independent result, the reported gains in stability and geographic compactness are not shown to be independent of the fitting process. No separate held-out validation or fixed-hyperparameter comparison is described. This matches the fitted-input-called-prediction pattern but is only partial because external data and standard clustering still supply the raw inputs.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 0 invented entities

The central claim rests on the assumption that the chosen variables and distance metric produce meaningful clusters and that the custom selection algorithm correctly identifies stable partitions; no new entities are postulated.

free parameters (1)
  • clustering hyperparameters (number of clusters, weighting of distance term)
    Chosen via the custom algorithm that balances homogeneity, separation, and geo versus non-geo comparison.
axioms (2)
  • domain assumption Ward-like linkage combined with geographic distances yields stable and interpretable partitions of awareness data.
    Invoked when claiming superiority of the geographically-informed version.
  • domain assumption The awareness, socio-economic, and climate variables are measured without systematic cross-country bias.
    Required for the clusters to reflect true awareness differences.

pith-pipeline@v0.9.0 · 5744 in / 1361 out tokens · 29234 ms · 2026-05-23T21:02:56.960605+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

34 extracted references · 34 canonical work pages · 1 internal anchor

  1. [1]

    The velocity of climate change

    Scott R Loarie, Philip B Duffy, Healy Hamilton, Gregory P Asner, Christopher B Field, and David D Ackerly. The velocity of climate change. Nature, 462(7276):1052–1055, 2009

    work page 2009

  2. [2]

    Climate change: Strategies for mitigation and adaptation

    Fang Wang, Jean Damascene Harindintwali, Ke Wei, Yuli Shan, Zhifu Mi, Mark John Costello, Sabine Grunwald, Zhaozhong Feng, Faming Wang, Yuming Guo, et al. Climate change: Strategies for mitigation and adaptation. The Innovation Geoscience, 1(1):100015–61, 2023

    work page 2023

  3. [3]

    Multi-decadal increase of forest burned area in australia is linked to climate change

    Josep G Canadell, CP Meyer, Garry D Cook, Andrew Dowdy, Peter R Briggs, Jürgen Knauer, Acacia Pepler, and Vanessa Haverd. Multi-decadal increase of forest burned area in australia is linked to climate change. Nature communications, 12(1):6921, 2021

    work page 2021

  4. [4]

    Climate change is increasing the likelihood of extreme autumn wildfire conditions across california

    Michael Goss, Daniel L Swain, John T Abatzoglou, Ali Sarhadi, Crystal A Kolden, A Park Williams, and Noah S Diffenbaugh. Climate change is increasing the likelihood of extreme autumn wildfire conditions across california. Environmental Research Letters, 15(9):094016, 2020

    work page 2020

  5. [5]

    Understanding human influence on climate change in china

    Ying Sun, Xuebin Zhang, Yihui Ding, Deliang Chen, Dahe Qin, and Panmao Zhai. Understanding human influence on climate change in china. National science review, 9(3):nwab113, 2022

    work page 2022

  6. [6]

    Drought risk for agricultural systems in south africa: Drivers, spatial patterns, and implications for drought risk management

    Isabel Meza, Ehsan Eyshi Rezaei, Stefan Siebert, Gohar Ghazaryan, Hamideh Nouri, Olena Dubovyk, Helena Gerdener, Claudia Herbert, Jürgen Kusche, Eklavyya Popat, et al. Drought risk for agricultural systems in south africa: Drivers, spatial patterns, and implications for drought risk management. Science of the Total Environment, 799:149505, 2021

    work page 2021

  7. [7]

    Recent responses to climate change reveal the drivers of species extinction and survival

    Cristian Román-Palacios and John J Wiens. Recent responses to climate change reveal the drivers of species extinction and survival. Proceedings of the National Academy of Sciences, 117(8):4211–4217, 2020

    work page 2020

  8. [8]

    A million threatened species? thirteen questions and answers, 2019

    Purvis Andy. A million threatened species? thirteen questions and answers, 2019. https://www.ipbes.net/ news/million-threatened-species-thirteen-questions-answers , Last accessed on August 31 2024

    work page 2019

  9. [9]

    Greenhouse effect, sea level rise, and coastal zone management

    James G Titus. Greenhouse effect, sea level rise, and coastal zone management. Coastal Management, 14(3):147– 171, 1986

    work page 1986

  10. [10]

    Ecological development and global climate change: A cross-national study of kyoto protocol ratification

    Sammy Zahran, Eunyi Kim, Xi Chen, and Mark Lubell. Ecological development and global climate change: A cross-national study of kyoto protocol ratification. Society and Natural Resources, 20(1):37–55, 2007

    work page 2007

  11. [11]

    What are the costs of limiting co2 concentrations

    James A Edmonds and Ronald D Sands. What are the costs of limiting co2 concentrations. Global climate change: The science, economics, and politics, pages 140–86, 2003

    work page 2003

  12. [12]

    Global climate risk index, 2021

    David Eckstein, Vera Künzel, and Schäfer Laura. Global climate risk index, 2021. https://reliefweb.int/ report/world/global-climate-risk-index-2021 , Last accessed on August 31 2024

    work page 2021

  13. [13]

    A survey of constrained classification

    AD Gordon. A survey of constrained classification. Computational Statistics & Data Analysis, 21(1):17–29, 1996

    work page 1996

  14. [14]

    Clustering of spatial data by the em algorithm

    Christophe Ambroise, Mo Dang, and Gérard Govaert. Clustering of spatial data by the em algorithm. In geoENV I—Geostatistics for Environmental Applications: Proceedings of the Geostatistics for Environmental Applications Workshop, Lisbon, Portugal, 18–19 November 1996, pages 493–504. Springer, 1997. 17 Clustering climate change awareness

    work page 1996

  15. [15]

    Clustering spatial data with a geographic constraint: exploring local search

    Zhung-Xun Liao and Wen-Chih Peng. Clustering spatial data with a geographic constraint: exploring local search. Knowledge and information systems, 31:153–170, 2012

    work page 2012

  16. [16]

    Spatially constrained clustering of ecological networks

    Vincent Miele, Franck Picard, and Stéphane Dray. Spatially constrained clustering of ecological networks. Methods in Ecology and Evolution, 5(8):771–779, 2014

    work page 2014

  17. [17]

    Constrained clustering of irregularly sampled spatial data

    Yudi Pawitan and Jian Huang. Constrained clustering of irregularly sampled spatial data. Journal of Statistical Computation and Simulation, 73(12):853–865, 2003

    work page 2003

  18. [18]

    A geostatistical basis for spatial weighting in multivariate classification

    MA Oliver and R Webster. A geostatistical basis for spatial weighting in multivariate classification. Mathematical geology, 21:15–35, 1989

    work page 1989

  19. [19]

    The multivariate (co) variogram as a spatial weighting function in classification methods

    Gilles Bourgault, Denis Marcotte, and Pierre Legendre. The multivariate (co) variogram as a spatial weighting function in classification methods. Mathematical Geology, 24:463–478, 1992

    work page 1992

  20. [20]

    Clustgeo: an r package for hierarchical clustering with spatial constraints

    Marie Chavent, Vanessa Kuentz-Simonet, Amaury Labenne, and Jérôme Saracco. Clustgeo: an r package for hierarchical clustering with spatial constraints. Computational Statistics, 33(4):1799–1822, 2018

    work page 2018

  21. [21]

    International public opinion on climate change

    Anthony Leiserowitz, Jennifer Carman, Nicole Buttermore, Xinran Wang, Seth Rosenthal, Jennifer Marlon, and Kelsey Mulcahy. International public opinion on climate change. New Haven, CT: Yale Program on Climate Change Communication and Facebook Data for Good, 2021

    work page 2021

  22. [22]

    International public opinion on climate change 2022

    Anthony Leiserowitz, Jennifer Carman, Nicole Buttermore, Liz Neyens, Seth Rosenthal, Jennifer Marlon, JW Schneider, and Kelsey Mulcahy. International public opinion on climate change 2022. New Haven, CT: Yale Program on Climate Change Communication and Facebook Data for Good, 2022

    work page 2022

  23. [23]

    R: A Language and Environment for Statistical Computing

    R Core Team. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria, 2023

    work page 2023

  24. [24]

    dendextend: an r package for visualizing, adjusting, and comparing trees of hierarchical clustering

    Tal Galili. dendextend: an r package for visualizing, adjusting, and comparing trees of hierarchical clustering. Bioinformatics, 2015

    work page 2015

  25. [25]

    Comparing clusterings and numbers of clusters by aggregation of calibrated clustering validity indexes

    Serhat Emre Akhanli and Christian Hennig. Comparing clusterings and numbers of clusters by aggregation of calibrated clustering validity indexes. Statistics and Computing, 30(5):1523–1544, 2020

    work page 2020

  26. [26]

    Wilderjans

    Julian Rossbroich, Jeffrey Durieux, and Tom F. Wilderjans. Model selection strategies for determining the optimal number of overlapping clusters in additive overlapping partitional clustering. Journal of Classification, 39(2):264–301, 2022

    work page 2022

  27. [27]

    Multidimensional spatiotemporal clustering -- An application to environmental sustainability scores in Europe

    Caterina Morelli, Simone Boccaletti, Paolo Maranzano, and Philipp Otto. Multidimensional spatiotemporal clustering–an application to environmental sustainability scores in europe. arXiv preprint arXiv:2405.20191, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  28. [28]

    Nbclust: An r package for determining the relevant number of clusters in a data set

    Malika Charrad, Nadia Ghazzali, Véronique Boiteau, and Azam Niknafs. Nbclust: An r package for determining the relevant number of clusters in a data set. Journal of Statistical Software, 61(6):1 – 36, 2014

    work page 2014

  29. [29]

    Silhouettes: a graphical aid to the interpretation and validation of cluster analysis

    Peter J Rousseeuw. Silhouettes: a graphical aid to the interpretation and validation of cluster analysis. Journal of computational and applied mathematics, 20:53–65, 1987

    work page 1987

  30. [30]

    Well-separated clusters and optimal fuzzy partitions

    Joseph C Dunn. Well-separated clusters and optimal fuzzy partitions. Journal of cybernetics, 4(1):95–104, 1974

    work page 1974

  31. [31]

    A general statistical framework for assessing categorical clustering in free recall

    Lawrence J Hubert and Joel R Levin. A general statistical framework for assessing categorical clustering in free recall. Psychological bulletin, 83(6):1072, 1976

    work page 1976

  32. [32]

    A dendrite method for cluster analysis

    Tadeusz Cali´nski and Jerzy Harabasz. A dendrite method for cluster analysis. Communications in Statistics-theory and Methods, 3(1):1–27, 1974

    work page 1974

  33. [33]

    Clustisz: A program to test for the quality of clustering of a set of objects

    John O McClain and Vithala R Rao. Clustisz: A program to test for the quality of clustering of a set of objects. Journal of Marketing Research, pages 456–460, 1975

    work page 1975

  34. [34]

    Classification

    Allan David Gordon. Classification. CRC Press, 1999. 18 Clustering climate change awareness A Main analysis results: spatial hierarchical clustering using low and medium-low climate change awareness information Figure 13: Optimal values of the mixing parameter α conditioning on K clusters (that is, α∗ K) obtained using the criterion by Morelli et al. (202...

    work page 1999