Clustering Procedure Research Articles

From A-to-Z review of clustering validation indices

Data clustering involves identifying latent similarities within a dataset and organizing them into clusters or groups. The outcomes of various clustering algorithms differ as they are susceptible to the intrinsic characteristics of the original dataset, including noise and dimensionality. The effectiveness of such clustering procedures directly impacts the homogeneity of clusters, underscoring the significance of evaluating algorithmic outcomes. Consequently, the assessment of clustering quality presents a significant and complex endeavor. A pivotal aspect affecting clustering validation is the cluster validity metric, which aids in determining the optimal number of clusters. The main goal of this study is to comprehensively review and explain the mathematical operation of internal and external cluster validity indices, but not all, to categorize these indices and to brainstorm suggestions for future advancement of clustering validation research. In addition, we review and evaluate the performance of internal and external clustering validation indices on the most common clustering algorithms, such as the evolutionary clustering algorithm star (ECA*). Finally, we suggest a classification framework for examining the functionality of both internal and external clustering validation measures regarding their ideal values, user-friendliness, responsiveness to input data, and appropriateness across various fields. This classification aids researchers in selecting the appropriate clustering validation measure to suit their specific requirements.

Clustering with Minimum Spanning Trees: How Good Can It Be?

AbstractMinimum spanning trees (MSTs) provide a convenient representation of datasets in numerous pattern recognition activities. Moreover, they are relatively fast to compute. In this paper, we quantify the extent to which they are meaningful in low-dimensional partitional data clustering tasks. By identifying the upper bounds for the agreement between the best (oracle) algorithm and the expert labels from a large battery of benchmark data, we discover that MST methods can be very competitive. Next, we review, study, extend, and generalise a few existing, state-of-the-art MST-based partitioning schemes. This leads to some new noteworthy approaches. Overall, the Genie and the information-theoretic methods often outperform the non-MST algorithms such as K-means, Gaussian mixtures, spectral clustering, Birch, density-based, and classical hierarchical agglomerative procedures. Nevertheless, we identify that there is still some room for improvement, and thus the development of novel algorithms is encouraged.

Biopsychosocial phenotypes in people with HIV in the CHARTER cohort.

Neuropsychiatric complications such as neurocognitive impairment and depression are common in people with HIV despite viral suppression on antiretroviral therapy, but these conditions are heterogeneous in their clinical presentations and associated disability. Identifying novel biopsychosocial phenotypes that account for neurocognitive performance and depressive and functional symptoms will better reflect the complexities encountered in clinical practice and may have pathological and therapeutic implications. We classified 1580 people with HIV based on 17 features, including 7 cognitive domains, 4 subscales of the Beck depression inventory-II, 5 components of the patient's assessment of own functioning inventory, and dependence in instrumental and basic activities of daily living. A two-stage clustering procedure consisting of dimension reduction with self-organizing maps and Mahalanobis distance-based k-means clustering algorithms was applied to cross-sectional data. Baseline demographic and clinical characteristics were compared between the phenotypes, and their prediction on the biopsychosocial phenotypes was evaluated using multinomial logistic regression. Four distinct phenotypes were identified. Participants in Phenotype 1 overall did well in all domains. Phenotype 2 had mild-to-moderate depressive symptoms and the most substance use disorders. Phenotype 3 had mild-to-moderate cognitive impairment, moderate depressive symptoms, and the worst daily functioning; they also had the highest proportion of females and non-HIV conditions that could affect cognition. Phenotype 4 had mild-to-moderate cognitive impairment but with relatively good mood, and daily functioning. Multivariable analysis showed that demographic characteristics, medical conditions, lifetime cocaine use disorder, triglycerides, and non-antiretroviral therapy medications were important variables associated with biopsychosocial phenotype. We found complex, multidimensional biopsychosocial profiles in people with HIV that were associated with different risk patterns. Future longitudinal work should determine the stability of these phenotypes, assess factors that influence transitions from one phenotype to another, and characterize their biological associations.

Jacobian-scaled K-means clustering for physics-informed segmentation of reacting flows

This work introduces Jacobian-scaled K-means (JSK-means) clustering, which is a physics-informed clustering strategy centered on the K-means framework. The method allows for the injection of underlying physical knowledge into the clustering procedure through a distance function modification: instead of leveraging conventional Euclidean distance vectors, the JSK-means procedure operates on distance vectors scaled by matrices obtained from dynamical system Jacobians evaluated at the cluster centroids. The goal of this work is to show how the JSK-means algorithm – without modifying the input dataset – produces clusters that capture regions of dynamical similarity, in that the clusters are redistributed towards high-sensitivity regions in phase space and are described by similarity in the source terms of samples instead of the samples themselves. The algorithm is demonstrated on a complex reacting flow simulation dataset (a channel detonation configuration), where the dynamics in the thermochemical composition space are known through the highly nonlinear and stiff Arrhenius-based chemical source terms. Interpretations of cluster partitions in both physical space and composition space reveal how JSK-means shifts clusters produced by standard K-means towards regions of high chemical sensitivity (e.g., towards regions of peak heat release rate near the detonation reaction zone). The findings presented here illustrate the benefits of utilizing Jacobian-scaled distances in clustering techniques, and the JSK-means method in particular displays promising potential for improving former partition-based modeling strategies in reacting flow (and other multi-physics) applications.

AN AUTOMATED APPROACH TO SELECTING SENTENCES FOR TEST CASE GENERATION

The modern field of education is characterized by the increasing use of multiple choice tests to assess students’ knowledge and skills. One of the common methods of selecting sentences for such tests is the application of textual data clustering procedures. In this study, a module for sentence selection was developed that includes three steps: preprocessing, sentence parameter computation, and clustering. However, an objective evaluation of the quality of the obtained clusters using the silhouette coefficient and Davis-Boldin index showed that the clustering model used did not give satisfactory results.

Оценка хозяйственных рисков при реализации технологий угледобычи в современных условиях недропользования

Introduction. This paper presents research in the field of methodology of assessment and monitoring of associated under ground coalmining risks in order to level their impact on the resulting technical and economic indicators of production and economic activities in modern conditions of subsoiluse. The functioning of technological systems of coal mining enterprises at the present stage of subsurface use occurs in a highly dynamic and highly variable functional environment with a fair degree of uncertainty, which is generated by a significant variety of parameters and characteristics of the initial mining, geological and mining engineering environment, which, in turn, generates the emergence of various groups and risk factors for subsurface use. The established procedures and methods of risk forecasting currently do not give satisfactory results due to the globalization of world markets and high volatility of prices for coal products. Leveling the negative degree of influence of the risk component on the resulting indicators – indicators of technical and economic activity can be achieved through the introduction of an operational system for assessing and monitoring various groups of risk factors. Materials and methods. For the purpose of evaluation, it is proposed to use the methodological and methodological foundations of the method called “vector norms”, which involves the use of Affine space, the constituent areas of compromises and the introduction of utility estimates. A complete model representation of the assessment and monitoring of the impact of various risk factors of subsurface use, taking into account their clustering and ranking, is presented. The algorithmic support of this method is reduced to the following components-iterations: 1. Formation of criteria-based assessment matrices of various factors of subsurface use risk groups. 2. Assignment of optima to evaluation criteria. 3. In order to ensure the legality of the use and implementation of mathematical procedures and operations, the reduction of multi-dimensional evaluation criteria to a relative form, taking into account the assigned optima. 4. Formation of the objective function of the “norm of the vector” (integral indicator) and the development of a procedure for selecting the optimal option. Since the use of the methodological and methodological foundations of the method called “vector norms” allows us to form quantitative estimates of the presented factors-types of risk of subsurface use, this makes it possible to implement the clustering procedure or ranking of a selected set of coal mining enterprises in conjunction with existing risk areas. Results. Quite high values of correlation coefficients were obtained with a significant closeness of the relationship of inversely proportional dependencies of generalizing integral indicators of the degree of risk and the level of achievement of planned indicators and development of production capacity with the level of operational profitability from the conduct of production and economic activities of coal mining enterprises with a directly proportional dependence on the operating costs of mining 1t of coal. Discussion. Сomplete model representation of the assessment and monitoring of the impact of various risk factors of subsurface use, taking into account their clustering and ranking, is presented. As a practical component of the research, variable series of areas of integral degree of risk in the functional environment of Kuzbass coal mines are presented. The use of software modules for correlation and regression analysis, taking into account three-year retrospective dynamics, made it possible to establish the legality and objectivity of the proposed methodological apparatus for integrated risk assessment of the implementation of coal mining technologies in modern conditions of subsurface use. Conclusion. The results of the implementation of the theoretical components of the methodology and model representations of the developed conceptual foundations for the integrated risk assessment of the implementation of coal mining technologies in modern conditions of subsurface use, taking into account five-year retrospective dynamics, showed that in the Kuznetsk basin there was a steady tendency to reduce the cluster of coal mines, which is characterized by high risks of carrying out production and economic activities in the presence of a disproportion of resource potential (the emergent part). The accompanying processes of formation of this trend are manifested in a quantitative increase in the functional units of the rating cluster of group 2B with an average resource potential and inherent moderate risk. It is within this cluster that strategies for prioritizing investments for the development and renewal of technological systems and structures of coal mining are being formed, since only they are able to provide the fastest possible return on invested funds. Resume. The article presents the results of research in the field of methodology for assessing and monitoring the risks associated with underground coal mining in order to level out their impact on the resulting technical and economic indicators of production and economic activity in modern conditions of subsoil use. The research results can be useful in the implementation of coal mining technology in modern conditions of subsoil use.

Standard precipitation-temperature index (SPTI) drought identification by fuzzy c-means methodology

Global warming and climate change impacts intensify hydrological cycle and consequently unprecedented drought and flood appear in different parts of the world. Meteorological drought assessments are widely evaluated by the concept of standardized precipitation index (SPI), which provides drought classification. Its application is based on the probabilistic standardization procedure, but in the literature, there is a confusion with the statistical standardization procedure. This paper provides distinctive differences between the two approaches and provides the application of a better method. As a novel approach, SPI classification is coupled with fuzzy clustering procedure, which provides drought evaluation procedure based on two variables jointly, precipitation and temperature, which is referred to as the standard precipitation-temperature index (SPTI). The final product is in the form of fuzzy c-means clustering in five clusters with exposition of annual drought membership degrees (MDs) for each cluster and resulting objective function. The application of the proposed fuzzy methodology is presented for the long-term annual precipitation and temperature records from New Jersey Statewide records.

Sustainable, Values-based Banking Model – Clustering Procedure for assessing its Convergence Pattern across European Banks Iustina

Sustainable, Values-based Banking Model – Clustering Procedure for assessing its Convergence Pattern across European Banks Iustina

Traffic flows time series in a flood-prone area: modeling and clustering on extreme values with a spatial constraint

Time series of traffic flows, extracted from mobile phone origin–destination data, are employed for monitoring people crowding and mobility in areas subject to flooding risk. By applying a vector autoregressive model with exogenous covariates combined with dynamic harmonic regression to such time series, we detected the presence of many extreme events in the residuals, which exhibit heavy-tailed distribution. For this reason, we propose a time series clustering procedure based on tail dependence which is suitable for data characterized by a spatial dimension, since objects’ geographical proximity is taken into account. The final aim is to obtain clusters of areas characterized by the common tendency to the manifestation of extreme events, that in this case study are represented by extremely high incoming traffic flows. The proposed method is applied to the Mandolossa, a strongly urbanized area located on the western outskirts of Brescia (northern Italy) which is subject to frequent flooding.

Sieci neuronowe do mapowania archeologicznych artefaktów litowych

An analytical non-destructive strategy to chemically characterize lithic artefacts has been developed. Around 100 archaeological lithic materials found in Neolithic-Chalcolithic sites in the Mediterranean region of the Iberian Peninsula and nowadays stored in different museums of the Valencian Community (Spain), were studied. The materials belong to different typologies of rock (diabase, sillimanite, ophite and amphibolite) and were analysed employing portable energy dispersive X-ray fluorescence spectroscopy (pXRF) directly in the rock surface. The obtained data were processed through neural networks protocol, specifically the so-called Kohonen networks or Self Organised Maps (SOM), to map the geologic samples. This self-organized topological feature maps are suitable to deal with multidimensional representations and map them in a two-dimensional space of neurons, following an unsupervised learning protocol. SOM is used to reduce multidimensional data onto lower-dimensional spaces and clustering procedures. As a result, SOM create spatially organized representations, which enhance the discovery of correlations between data. In this case the method has enabled the evaluation of elemental markers related to each rock type behaving as a fine hidden pattern detector and so understand the possible advantages and disadvantages of the analytical method employed to define provenance issues. The attribution suggested by statistics is mainly driven by the composition of rocks essential minerals which are linked to the different petrogenetic conditions. The results showed that in most of the cases the distribution and dispersion of the chemical profile depend of the kind of rock, and clearly suggest that a good way to identify stone tools raw material procurement is to look for elemental markers, being the prior step to create an approximation to ancient exchange networks and their evolution in a diachronic axis.

A Classification Method of Earthquake Ground Motion Records Based on the Results of K-Means Clustering Analysis

This paper presents a classification method for earthquake ground motion records utilizing the results of K-means cluster analysis. The moment magnitude and Joyner–Boore distance are utilized as the primary parameters for clustering the earthquake ground motion records. The classification boundaries are established through an examination of moment magnitude ranges, Joyner–Boore distance ranges, and spectral characteristics within each cluster. In this study, a comprehensive dataset comprising 7627 horizontal earthquake acceleration records was meticulously curated for analysis. The data were subjected to separate clustering and grouping procedures, allowing for insightful comparisons between the resultant clusters. Significant disparities in spectral characteristics across the classification groups were demonstrated. These differences become particularly pronounced when a moment magnitude threshold of 6 and a Joyner–Boore distance threshold of 140 km are employed to categorize the ground motion records. The approach underscores the substantial impact of classification based on earthquake ground motion spectral characteristics, while also mitigating the potential instabilities inherent in cluster analysis results. A refined and quantitatively robust framework for understanding and categorizing earthquake ground motions is provided, offering valuable insights for seismic data analysis and contributing to more accurate and reliable assessments of seismic activity.

Functional zoning of biodiversity profiles

AbstractSpatial mapping of biodiversity is crucial to investigate spatial variations in natural communities. Several indices have been proposed in the literature to represent biodiversity as a single statistic. However, these indices only provide information on individual dimensions of biodiversity, thus failing to grasp its complexity comprehensively. Consequently, relying solely on these single indices can lead to misleading conclusions about the actual state of biodiversity. In this work, we focus on biodiversity profiles, which provide a more flexible framework to express biodiversity through nonnegative and convex curves, which can be analyzed by means of functional data analysis. By treating the whole curves as single entities, we propose to achieve a functional zoning of the region of interest by means of a penalized model‐based clustering procedure. This provides a spatial clustering of the biodiversity profiles, which is useful for policy‐makers both for conserving and managing natural resources and revealing patterns of interest. Our approach is evaluated using a simulation study and discussed through the analysis of the Harvard Forest Data, which provides information on the spatial distribution of woody stems within a plot of the Harvard Forest.

Parsing evoked and induced gamma response differences in Autism: A visual evoked potential study

ObjectiveElectroencephalography (EEG) measures of visual evoked potentials (VEPs) provide a targeted approach for investigating neural circuit dynamics. This study separately analyses phase-locked (evoked) and non-phase-locked (induced) gamma responses within the VEP to comprehensively investigate circuit differences in autism. MethodsWe analyzed VEP data from 237 autistic and 114 typically developing (TD) children aged 6–11, collected through the Autism Biomarkers Consortium for Clinical Trials (ABC-CT). Evoked and induced gamma (30–90 Hz) responses were separately quantified using a wavelet-based time–frequency analysis, and group differences were evaluated using a permutation-based clustering procedure. ResultsAutistic children exhibited reduced evoked gamma power but increased induced gamma power compared to TD peers. Group differences in induced responses showed the most prominent effect size and remained statistically significant after excluding outliers. ConclusionsOur study corroborates recent research indicating diminished evoked gamma responses in children with autism. Additionally, we observed a pronounced increase in induced power. Building upon existing ABC-CT findings, these results highlight the potential to detect variations in gamma-related neural activity, despite the absence of significant group differences in time-domain VEP components. SignificanceThe contrasting patterns of decreased evoked and increased induced gamma activity in autistic children suggest that a combination of different EEG metrics may provide a clearer characterization of autism-related circuitry than individual markers alone.

Geo-constrained clustering of resistivity data revealing the heterogeneous lithological architectures and the distinctive geoelectrical signature of shallow deposits

For all applications, subsurface models should be consistent with all available geological and geophysical knowledge. Current practices for synergistic interpretation of geological and geophysical approaches often rely on purely qualitative comparisons, resulting sometimes in inconsistent findings. This study introduces a procedure for a statistical and geo-constrained clustering of electrical resistivity data derived from Electrical Resistivity Tomography (ERT) to address this gap, providing a quantitative parameterization for site-specific geoelectrical signatures of litho-stratigraphic architectures. Seventeen boreholes and three ERT surface profiles were employed to link geophysical inversion results to geological criteria. Core samples allowed grain size analyses, while geological-statistical clustering of electrical resistivity, driven by the observation of stratigraphic contacts in drilled boreholes, established a parametric relationship between geology and geophysics. The iterative clustering procedure, utilizing a classification algorithm, geological boundary constraints, and granulometric analyses, discriminated six distinct lithological clusters, capturing the lateral and vertical heterogeneity of shallow deposits. Subsequent spatial grouping of anthropogenic materials delineated lithological structures and facilitated the classification and identification of filling materials, silty sands, clayey sands, and clays and silts, each exhibiting distinct resistivity variations. The geo-driven geophysical clustering revealed complex lithological structures, especially paleo-channels, capturing their unique geoelectric footprints. The iterative clustering of geo-constrained resistivity data emerges as a powerful tool for subsurface exploration, contributing significantly to understanding lithological heterogeneity, quantifying statistically-based geoelectrical parametrization of shallow sediments, and evaluating the resistivity signature of different deposits. By bridging the gap between geology and geophysics, this data-driven approach establishes a benchmark for future applications. For instance, in the context of contaminated sites, it can be applied to identify pollutants versus geological heterogeneities.

Cognitive Alterations Associated with Remission and Alcohol Dependence Severity in Ethnically Diverse Patients of Siberia.

Impaired cognition in individuals with alcohol dependence may be associated with increased relapse risk. It has been recorded in more than half of patients during six months after treatment. In certain ethnic groups, for example, Tuvinians, the indigenous people of Siberia, relapses occur in extremely short periods of one to three months after treatment. An approach currently used to alcohol dependence treatment may be less effective for these patients. The study aimed to investigate cognitive sequelae in indigenous Tuvinian patients with alcohol dependence. The sample included 166 patients, 74 of indigenous ethnicity (Tuvinians) and 92 non-indigenous white patients. Data on inhibitory control, cognitive flexibility, attention, and working memory were collected from all the patients and processed using cluster analysis. The clustering data were then complemented by indicators of disorder dynamics, impulsivity, and emotion regulation. The clustering procedure revealed groups with severe cognitive sequelae. More than four-fold attention decrease was found in 43.5% of non-indigenous patients, and more impaired cognitive flexibility was revealed among 60.8% of indigenous patients. Groups with severe cognitive sequelae had higher impulsivity, maladaptive emotion regulation, more hospitalizations, faster disease progression, and shorter remissions. The latter was significantly reduced to 90 days on average in the severe group of indigenous patients versus 135 days of remission in the non-indigenous severe group. Results obtained may advance tailored intervention in alcohol-dependent patients of the indigenous Tuvinian ethnicity. While little is still known about the alcohol dependence course and consequences in the indigenous Tuvinians of Siberia, this study contributes to the global mental health data on alcohol abuse and dependence in indigenous communities.

Clustering single-cell RNA sequencing data via iterative smoothing and self-supervised discriminative embedding.

Single-cell transcriptome sequencing (scRNA-seq) is a high-throughput technique used to study gene expression at the single-cell level. Clustering analysis is a commonly used method in scRNA-seq data analysis, helping researchers identify cell types and uncover interactions between cells. However, the choice of a robust similarity metric in the clustering procedure is still an open challenge due to the complex underlying structures of the data and the inherent noise in data acquisition. Here, we propose a deep clustering method for scRNA-seq data called scRISE (scRNA-seq Iterative Smoothing and self-supervised discriminative Embedding model) to resolve this challenge. The model consists of two main modules: an iterative smoothing module based on graph autoencoders designed to denoise the data and refine the pairwise similarity in turn to gradually incorporate cell structural features and enrich the data information; and a self-supervised discriminative embedding module with adaptive similarity threshold for partitioning samples into correct clusters. Our approach has shown improved quality of data representation and clustering on seventeen scRNA-seq datasets against a number of state-of-the-art deep learning clustering methods. Furthermore, utilizing the scRISE method in biological analysis against the HNSCC dataset has unveiled 62 informative genes, highlighting their potential roles as therapeutic targets and biomarkers.

Digital Empowerment: How Small B to B Companies Remodel the Value Chain Through Digital Platforms

ABSTRACT Purpose Digital platforms represent a major technological change that affects the entire value chain of a company and requires digital resources and a flexible organizational structure. Given the resource limitations that SMEs have, it is interesting to find out to what extent they have adopted digital platforms. The aim of this paper is to explore the level of SMEs’ digitalization and how sophisticated these businesses are in using various digital platforms in primary and secondary value creation activities. Additionally, we set out to identify clusters with different needs and behaviors when considering digital platforms. Methodology/Approach To address these research questions, we conducted a survey using a quota sampling procedure to reflect the SMEs’ structure in size and industry. We analyzed the data collected from 194 respondents using a two-step clustering procedure to identify different groups of SMEs with respect to the use of digital platforms. For business activities, we used Porter’s value chain and for assessing the adoption rate, we applied the Moore and Benbasat scale to measure adopters’ perceptions of the technology. Findings Most SMEs use general digital platforms while less than 20% have developed their own platforms or use the platforms of business partners. Organizations with a hybrid structure use the largest number of digital platforms, while enterprises with a fluid structure use the fewest. Companies are using their own platforms or partners’ platforms for primary activities while for support activities are employing general platforms. The most important factor which influences the adoption rate of digital platforms is the image and then comes the advantages they bring. Regarding the clusters, four distinct groups were identified with specific value creation behaviors Digital Pathfinders, Basic Techies, Tech Blend, and Flexible Adopters. Practical implications Practical implications concern both the companies that use digital platforms and the companies that develop them. From the user perspective, this study sets a benchmark for different industries and presents the most common practices Regarding the companies that develop digital platforms, it is important to know the main deterrence factors, namely the lack of testing possibilities and the difficulty in choosing the most suitable solution. Taking into account the specific needs of the four clusters, developers can adapt their product and value proposition to stimulate the adoption of the digital platform. Originality/Value/Contribution This paper contributes to research on digital platforms in the context of transformational entrepreneurship related to crisis or catastrophic events such as pandemics or wars. Furthermore, it adds to existing literature on digital transformation by providing recommendations considering specific value chain activities. Although there are numerous studies on digital platforms, most of them come from developed countries, which makes them less relevant for developing/underdeveloped ones.

Non-convex tensorial multi-view clustering by integrating [formula omitted]-based sliced-Laplacian regularization and [formula omitted]-sparsity

Consider the recent upswing in interest around multi-view clustering procedures. Such methods aim to boost clustering efficiency by leveraging information from numerous perspectives. Much research has been devoted to tensorial representation to exploit high-order correlations underlying disparate views while preserving the local geometric structure inside each view. Our research introduces a novel multi-view clustering approach. This approach creates a 3rd-order tensor, assimilating features from all perspectives. We use the t-product in the tensor space to generate the self-representation tensor from the tensorial data. We incorporate the ℓ1-based sliced-Laplacian regularization to increase our model’s resilience and introduce a fresh column-wise sparse norm: the ℓ2,p-norm with 0<p<1. This norm displays attributes of invariance, continuity, and differentiability. We present a closed-form answer to the ℓ2,p-regularized shrinkage problem, broadening its relevance to other generalized problems. Simultaneously, we propose a tensorial arctan-function as an improved surrogate for the tensor rank. This function has proven more proficient at assessing consistency across multiple viewpoints. By integrating these two components, we formulate an effective algorithm that refines our suggested model, ensuring that the constructed sequence gravitates toward the stationary KKT point. Our team conducts extensive experiments on various datasets to evaluate our model’s effectiveness, spanning diverse situations and scales. Results from these experiments emphasize that our approach establishes a novel performance standard.

Connectivity-Based Clustering for Mixed Discrete and Continuous Data

This paper introduces a density-based clustering procedure for datasets with variables of mixed type. The proposed procedure, which is closely related to the concept of shared neighbourhoods, works particularly well in cases where the individual clusters differ greatly in terms of the average pairwise distance of the associated objects. Using a number of concrete examples, it is shown that the proposed clustering algorithm succeeds in allowing the identification of subgroups of objects with statistically significant distributional characteristics.

Climate model selection via conformal clustering of spatial functional data

Climate model selection stands as a critical process in climate science and research. It involves choosing the most appropriate climate models to address specific research questions, simulating climate behaviour, or making projections about future climate conditions. This paper proposes a new approach, using spatial functional data analysis, to asses which of the 18 EURO CORDEX simulation models work better for predicting average temperatures in the Campania region (Italy). The method involves two key steps: first, using functional data analysis to process climate variables and select optimal models by a hierarchical clustering procedure; second, validating the chosen models by proposing a new conformal prediction approach to the anomalies associated to each cluster.