Cluster Analysis Of Time Series Research Articles

Review of Clustering Methods for Functional Data

Functional data clustering is to identify heterogeneous morphological patterns in the continuous functions underlying the discrete measurements/observations. Application of functional data clustering has appeared in many publications across various fields of sciences, including but not limited to biology, (bio)chemistry, engineering, environmental science, medical science, psychology, social science, and so on. The phenomenal growth of the application of functional data clustering indicates the urgent need for a systematic approach to develop efficient clustering methods and scalable algorithmic implementations. On the other hand, there is abundant literature on the cluster analysis of time series, trajectory data, spatio-temporal data, and so on, which are all related to functional data. Therefore, an overarching structure of existing functional data clustering methods will enable the cross-pollination of ideas across various research fields. We here conduct a comprehensive review of original clustering methods for functional data. We propose a systematic taxonomy that explores the connections and differences among the existing functional data clustering methods and relates them to the conventional multivariate clustering methods. The structure of the taxonomy is built on three main attributes of a functional data clustering method and therefore is more reliable than existing categorizations. The review aims to bridge the gap between the functional data analysis community and the clustering community and to generate new principles for functional data clustering.

FORECASTING GREENHOUSE GAS EMISSIONS IN THE INDUSTRIAL PRODUCTION OF THE REPUBLIC OF KAZAKHSTAN

Excessive greenhouse gas (GHG) emissions are an environmental problem. Studies to determine cost-effective ways to reduce GHG emissions have revealed the need to model the dynamics of emissions of carbon dioxide, nitrous oxide, methane, and other gases. In this study, the calculation of CO2 equivalent emissions from industrial processes and production in the territory of the Republic of Kazakhstan was carried out. When forecasting, the data provided by the UN Framework Convention on Climate Change were used. To predict CO2 emissions from industrial production, tools for analysis and forecasting of time series were used: Prophet method, Cluster analysis of k-means time series, modern versions of ARIMA algorithms, exponential smoothing methods, and linear regression. This study presents comparative simulation results based on a baseline scenario with no action until 2045.This study compares four models to suggest an effective one for future CO2 emission forecasting. The accuracy comparison is conducted using various error measures, with the mean absolute percentage error (MAPE) chosen as the metric for comparison.

Geographic Variations in Human Mobility Patterns during the First Six Months of the COVID-19 Pandemic in California

Human mobility influenced the spread of the COVID-19 virus, as revealed by the high spatiotemporal granularity location service data gathered from smart devices. We conducted time series clustering analysis to delineate the relationships between human mobility patterns (HMPs) and their social determinants in California (CA) using aggregated smart device tracking data from SafeGraph. We first identified four types of temporal patterns for five human mobility indicator changes by applying dynamic-time-warping self-organizing map clustering methods. We then performed an analysis of variance and linear discriminant analysis on the HMPs with 17 social, economic, and demographic variables. Asians, children under five, adults over 65, and individuals living below the poverty line were found to be among the top contributors to the HMPs, including the HMP with a significant increase in the median home dwelling time and the HMP with emerging weekly patterns in full-time and part-time work devices. Our findings show that the CA shelter-in-place policy had varying impacts on HMPs, with socially disadvantaged places showing less compliance. The HMPs may help practitioners to anticipate the efficacy of non-pharmaceutical interventions on cases and deaths in pandemics.

Multivariate Time Series Clustering of Groundwater Quality Data to Develop Data-Driven Monitoring Strategies in a Historically Contaminated Urban Area

As groundwater quality monitoring networks have been expanded over the last decades, significant time series are now available. Therefore, a scientific effort is needed to explore innovative techniques for groundwater quality time series exploitation. In this work, time series exploratory analysis and time series cluster analysis are applied to groundwater contamination data with the aim of developing data-driven monitoring strategies. The study area is an urban area characterized by several superimposing historical contamination sources and a complex hydrogeological setting. A multivariate time series cluster analysis was performed on PCE and TCE concentrations data over a 10 years time span. The time series clustering was performed based on the Dynamic Time Warping method. The results of the clustering identified 3 clusters associated with diffuse background contamination and 7 clusters associated with local hotspots, characterized by specific time profiles. Similarly, a univariate time series cluster analysis was applied to Cr(VI) data, identifying 3 background clusters and 7 hotspots, including 4 singletons. The clustering outputs provided the basis for the implementation of data-driven monitoring strategies and early warning systems. For the clusters associated with diffuse background contaminations and those with constant trends, trigger levels were calculated with the 95° percentile, constituting future threshold values for early warnings. For the clusters with pluriannual trends, either oscillatory or monotonous, specific monitoring strategies were proposed based on trends’ directions. Results show that the spatio-temporal overview of the data variability obtained from the time series cluster analysis helped to extract relevant information from the data while neglecting measurements noise and uncertainty, supporting the implementation of a more efficient groundwater quality monitoring.

Combining Remote Sensing and Space-Time Analysis for Desertification Monitoring in the Semiarid Dryland of Nigeria

Desertification has been identified as the resultant effect of dryland loss. Desertification is catalysed by anthropogenic modifications and variations in environmental/climatic conditions. The situation in Nigeria is further exacerbated by the growing demand for land by the population. To this effect, this study performed a space-time analysis of vegetative cover between 2001 and 2020 to unravel patterns and trends across the semiarid region of the dryland system in Nigeria. The dynamics during the rainy season (May and September) were examined using the Terra Moderate Resolution Imaging Spectroradiometer (MODIS) dataset subjected to space-time analysis. Generalised Difference Vegetation Index (GDVI) was computed to the power of 2 to quantify vegetative cover across the study area. The results showed that the average of the GDVI ranges between -0.40 and 0.94, with a standard deviation of 0.11. Time series cluster analysis revealed that there are two temporal clusters: (1) no statistically significant trend (Statistics= 1.33, p-value = 0.18) and (2) statistically significant downtrend (Statistics = -2.37, p=0.02), with cluster 1 covering 95% of the areas examined. The most dominant (97% of the area) emerging space-time pattern was cold-spots (persistent, diminishing, sporadic, oscillating, and historical types). In conclusion, most of the areas showed no definite temporal pattern of vegetation pattern during the period, while more than 90% of the areas have witnessed a decline in vegetative cover. There is a need for a more coordinated approach to desertification control, constant monitoring is pertinent while new approaches to restoring degraded land are recommended.

Exploring dynamic change in arterial base excess with patient outcome and lactate clearance in the intensive care unit by hierarchical time-series clustering.

Hyperlactatemia is common in the intensive care unit (ICU) and relevant to prognosis, while the process of lactate normalization requires a relatively long period. We hypothesized that the dynamic change in base excess (BE) would be associated with ICU mortality and lactate clearance. We performed a retrospective cohort study of adult patients with hyperlactatemia admitted to the ICU from 2016 to 2021. The patients were divided into two groups according to whether the peak BE in 12 h was reached in the first 6 h. We compared ICU mortality and lactate clearance at 6 and 12 h after ICU admission. During the study period, 1,608 patients were admitted to the ICU with a lactate concentration of >2.0 mmol/L and stayed in the ICU for >24 h. The mortality rate was 11.2%. The patients were divided into two groups according to whether the peak BE was reached in the first 6 h following ICU admission: Peak BE12h ≤ 6h and Peak BE12h > 6h. The patients were also recorded as whether bicarbonate treatment was received (bicarbonate group, CRRT included) or not (non-bicarbonate group). Furthermore, lactic acid clearance patterns were identified by time-series clustering (TSC) using various algorithms and distance measures. We compared ICU mortality and lactate clearance at 6 and 12 h after ICU admission with logistic regression. After adjustment for other confounding factors, we found that Peak BE12h > 6h was independently associated with ICU mortality with an odds ratio of 2.231 (p = 0.036) in the bicarbonate group and 2.359 (p < 0.005) in the non-bicarbonate group. In addition, based on the definition of >10% lactate clearance at 6 h or >30% at 12 h, we found that Peak BE12h ≤ 6h had 85.2% sensitivity and 38.1% specificity for effective lactate clearance. In time-series clustering analysis, four categories were discriminated, and pattern of lactic acid clearance reveals the early prognostic value of BE in clearance of lactic acid. A prolonged time to reaching the peak BE was independently associated with ICU mortality. In patients with hyperlactatemia, Peak BE12h ≤ 6h could be used as an indicator to predict effective lactate clearance.

Regional Economic Disparities in Europe: Time-Series Clustering of NUTS 3 Regions

The aim of this research is to identify the regional economic disparities in the level of economic wealth and its dynamics in the NUTS 3 regions in EU28 over the period from 2000 to 2017. By performing a time-series clustering analysis at NUTS 3 level, we expect to uncover the economic disparities that might have been hidden in the aggregate NUTS 2 regions. Our results indicate that at a finer spatial scale (NUTS 3 level) disparities flourish, particularly in the period after the global crisis of 2008, in which different recovery rates are observed. In general, NUTS 2 regions tend to spatially cluster at the national level and, although NUTS 3 regions show slightly this tendency as well, the spatial effect is not as strong as it is for NUTS 2 level, revealing specific behaviours of the local economies and markets that remain hidden at the aggregate NUTS 2 level.

Simulating the effects of mobility restrictions in the spread of SARS-CoV-2 in metropolitan areas in Portugal.

Commuting flows and long-distance travel are important spreading factors of viruses and particularly airborne ones. Therefore, it is relevant to examine the association among diverse mobility scenarios and the spatial dissemination of SARS-CoV-2 cases. We intended to analyze the patterns of virus spreading linked to different mobility scenarios, in order to better comprehend the effect of the lockdown measures, and how such measures can be better informed. We simulated the effects of mobility restrictions in the spread of SARS-CoV-2 amongst the municipalities of two metropolitan areas, Lisbon (LMA) and Porto (PMA). Based on an adapted SEIR (Suscetible-Exposed-Infected-Removed) model, we estimated the number of new daily infections during one year, according to different mobility scenarios: restricted to essential activities, industrial activities, public transport use, and a scenario with unrestricted mobility including all transport modes. The trends of new daily infections were further explored using time-series clustering analysis, using dynamic time warping. Mobility restrictions resulted in lower numbers of new daily infections when compared to the unrestricted mobility scenario, in both metropolitan areas. Between March and September 2020, the official number of new infections followed overall a similar timeline to the one simulated considering only essential activities. At the municipal level, trends differ amongst the two metropolitan areas. The analysis of the effects of mobility in virus spread within different municipalities and regions could help tailoring future strategies and increase the public acceptance of eventual restrictions.

Landsat 8 영상을 이용한 산불피해지 시계열변화 분석방법에 관한 연구

Forests can be affected by wildfire in the form of primary damage (direct burn) and secondary damage (indirect withering). The distribution of the primary and secondary damages can change according to the elapsed time after a wildfire. This paper examined the method to analyze the temporal change of the damaged areas by wildfire. We proposed a time-series cluster analysis to categorize the wildfire-damaged areas in terms of the temporal change characteristics. A case study of the Kangwon fire in March 2022 shows that the damaged areas were divided into several categories, and three out of the nine groups were affected by the secondary damage. The time-series cluster map can provide information about the areas of the severe damage right after the wildfire and the subsequent damage one or more months later.

Towards a new urban geography of expenditure: Using bank card transactions data to analyze multi-sector spatiotemporal distributions

The spatial distribution of commercial activities is vital to support healthy lifestyles and to achieve livable public spaces and environmental, social and economic sustainability in our cities. However, commercial activities require a constant flow of expenditure for their own viability. As a result, understanding the spatial and temporal distribution of expenditure is fundamental, although the lack of detailed, complete data sources has impeded this task until now.Bank card data paves the way for a new urban geography of expenditure, thanks to its fine spatial and temporal granularity along with the uniform coverage of all commercial sectors. In this paper, we analyze temporal, spatial, and spatiotemporal distributions of expenditure at the intraurban scale of the city of Madrid (Spain), combining spatial statistical tools (Getis-Ord General for global autocorrelation and Getis-Ord Gi* hot spot analysis for local autocorrelation) with k-means cluster analysis and spatiotemporal tools (Time Series Clustering analysis and Temporal Hot Spot Analysis).Our analysis confirms the strong center-periphery gradient described in previous literature, but with a CBD integrated by distinct specialized areas. The paper demonstrates that bank card data has a great potential to support a new geography of expenditure that could strengthen decision-making in planning and retailing.

Time series clustering analysis of genes during osteogenic differentiation of human mesenchymal stem cells.

To investigate the gene expression pattern and related biological changes during osteogenic differentiation of human mesenchymal stem cells (hMSCs), we downloaded expression data for four uninduced hMSC samples, and 12 osteogenic induction samples at day 2, 8, 12 or 25, in the GSE37558 dataset. Differentially expressed genes (DEGs) between groups were screened, followed by short time-series expression miner (STEM) analysis and weighted gene co-expression network analysis (WGCNA). Osteogenic differentiation-related genes were extracted from the GeneCards database. Next, functional enrichment was performed, and protein-protein interaction (PPI) and lncRNA-miRNA-mRNA networks were constructed. Compared to uninduced hMSC samples, 163, 341, 447 and 537 DEGs were found in osteogenic induction samples at day 2, 8, 12 and 25, respectively, showing a sustainably increased trend. From STEM, WGCNA and the GeneCards database, a total of 107 key genes associated with osteogenic differentiation were screened; these genes were enriched in biological processes, such as ossification, ECM-receptor interaction, vasculature development, cartilage development and bone mineralization, as well as the Wnt signaling pathway and the chemokine signaling pathway. The PPI network identified four hub genes, STAT5A, TWIST1, FOXO1 and LEP. The lncRNA-miRNA-mRNA network revealed regulatory axes for STAT5A, FOXO1 and LEP. Three and two regulatory axes were found for STAT5A and LEP, respectively. Multiple regulatory axes for FOXO1 were found, such as MIR155HG-miR-223-FOXO1. This study identifies candidate key targets that may play important roles in regulating osteogenic differentiation of hMSCs, and provides novel information to further investigate the molecular regulation mechanism. More experiments are required to evaluate the effects of these genes on osteogenic differentiation of hMSCs.

Data-driven decision management of urban underground infrastructure through groundwater-level time-series cluster analysis: the case of Milan (Italy)

The significant increase in urbanization has resulted in greater use of the subsurface in urban planning and, therefore, increased interaction between groundwater and underground infrastructure. Numerical models are the primary tool adopted to manage the resulting problems; however, their construction is time- and cost-consuming. Groundwater-level time-series analysis can be a complementary method, as this data-driven approach does not require an extensive understanding of the geological and boundary conditions, even if providing insights into the hydrogeologic behaviour. Thus, a data-driven approach was adopted to analyse groundwater time-series of the shallow aquifer, occupied by several underground structures, beneath Milan city (Northern Italy). Statistical (Mann-Kendall and Sen’s slope estimator, autocorrelation and cross-correlation, hierarchical cluster analysis) and geospatial techniques were used to detect the potential variables influencing the groundwater levels of 95 monitoring wells, covering the period 2005–2019. A general rising trend of the water table was identified, with local hydrogeologic differences in the western and southernmost areas. Based on time-series analysis results, four management areas have been identified. These areas could act as future geographic units with specific groundwater management strategies. In particular, subsurface public car parks can be classified with respect to groundwater flooding as (1) not submerged, (2) possibly critical, or (3) submerged at different groundwater conditions. According to these outcomes, targeted guidelines for constructing new car parks have been elaborated for each management area. The methodology proved to be efficient in improving the urban conceptual model and helping stakeholders design the planned underground development, considering groundwater aspects.

Cluster and Trend Analysis of Rainfall Time Series in the Nakdong River Basin

In Korea, it is difficult to efficiently manage water resources, due to the variability of rainfall in addition to the increasing outflow due to climate change in recent years. Therefore, if the trends and characteristics of rainfall at each station can be identified in advance, the problems caused by the variability of rainfall can be effectively dealt with. In this study, the data on rainfall characteristics of 64 rainfall stations in the Nakdong river basin were collected for the period 2000 to 2019. The data were analyzed according to the elevation of each station using K-means cluster analysis, and the rainfall trends at each station were identified by homogeneity test and modified Mann-Kendall test. The analysis showed an increasing trend in March, April, November, December, spring and autumn, and a decreasing trend in January, May, September, summer and year. Also, based on the cluster analysis, it was confirmed that when the number of clusters was set to three, the rainfall characteristics were different depending on the elevation of each station. It is believed that linking the characteristics of rainfall by cluster and the results of trend analysis by station derived from the study can be used to come up with a water resource management plan that takes into account the variability of rainfall.

The Collapse of Infectious Disease Diagnoses Commonly Due to Communicable Respiratory Pathogens During the Coronavirus Disease 2019 Pandemic: A Time Series and Hierarchical Clustering Analysis.

BackgroundNonpharmaceutical interventions such as physical distancing and mandatory masking were adopted in many jurisdictions during the coronavirus disease 2019 pandemic to decrease spread of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). We determined the effects of these interventions on incidence of healthcare utilization for other infectious diseases.MethodsUsing a healthcare administrative dataset, we employed an interrupted time series analysis to measure changes in healthcare visits for various infectious diseases across the province of Ontario, Canada, from January 2017 to December 2020. We used a hierarchical clustering algorithm to group diagnoses that demonstrated similar patterns of change through the pandemic months.ResultsWe found that visits for infectious diseases commonly caused by communicable respiratory pathogens (eg, acute bronchitis, acute sinusitis) formed distinct clusters from diagnoses that often originate from pathogens derived from the patient’s own flora (eg, urinary tract infection, cellulitis). Moreover, infectious diagnoses commonly arising from communicable respiratory pathogens (hierarchical cluster 1: highly impacted diagnoses) were significantly decreased, with a rate ratio (RR) of 0.35 (95% confidence interval [CI], .30–.40; P < .001) after the introduction of public health interventions in April–December 2020, whereas infections typically arising from the patient’s own flora (hierarchical cluster 3: minimally impacted diagnoses) did not demonstrate a sustained change in incidence (RR, 0.95 [95% CI, .90–1.01]; P = .085).ConclusionsPublic health measures to curtail the incidence of SARS-CoV-2 were widely effective against other communicable respiratory infectious diseases with similar modes of transmission but had little effect on infectious diseases not strongly dependent on person-to-person transmission.

Social Influence Network Simulation Design Affects Behavior of Aggregated Entropy

As agents interact and influence one another in a social network, the opinions they hold about some common topic can change over time. These changes may enable us to infer mechanisms of the network that control how interactions lead to opinion change. Inferring such mechanisms from opinion data could enable analysis of social influence in data-sparse scenarios. However, limited work has focused on this problem, despite its clear value. To address this gap, we create opinion data using agent-based simulation and experimental design. By viewing opinion changes as an information-generating process, opinion dynamics can be studied using entropy. This work explores the relationships between aggregated entropy and five simulation design factors. Three entropy measures are calculated on continuous-valued opinions and are analyzed using a main effects model and cluster analysis. Overall, the choices of influence model and error distribution are most important to the entropy measures, activation regime is important to some measures, and population size is unimportant. Also, design variation can be detected using time-series cluster analysis. These findings may support work in inferring properties about real-world social influence networks using opinion data collected from their members.

Single-cell imaging of ERK and Akt activation dynamics and heterogeneity induced by G-protein-coupled receptors.

ABSTRACTKinases play key roles in signaling networks that are activated by G-protein-coupled receptors (GPCRs). Kinase activities are generally inferred from cell lysates, hiding cell-to-cell variability. To study the dynamics and heterogeneity of ERK and Akt proteins, we employed high-content biosensor imaging with kinase translocation reporters. The kinases were activated with GPCR ligands. We observed ligand concentration-dependent response kinetics to histamine, α2-adrenergic and S1P receptor stimulation. By using G-protein inhibitors, we observed that Gq mediated the ERK and Akt responses to histamine. In contrast, Gi was necessary for ERK and Akt activation in response to α2-adrenergic receptor activation. ERK and Akt were also strongly activated by S1P, showing high heterogeneity at the single-cell level, especially for ERK. Cluster analysis of time series derived from 68,000 cells obtained under the different conditions revealed several distinct populations of cells that display similar response dynamics. ERK response dynamics to S1P showed high heterogeneity, which was reduced by the inhibition of Gi. To conclude, we have set up an imaging and analysis strategy that reveals substantial cell-to-cell heterogeneity in kinase activity driven by GPCRs.

The dependence of hydropower planning in relation to the influence of climate in Northeast Brazil

Water scarcity in Northeast Brazil has caused latent perturbations in hydropower generation, which is undesirable for Energy Planning. Thus, this study aims to identify the influence of climate on hydropower generation by Sobradinho Dam in Bahia by: (i) assessing the streamflow climatology (1964-2017) and rainfall (1964-2015) through time series analysis, hypothesis testing and cluster analysis; (ii) assessing hydropower generation (2000-2017) using climate and energy data, through principal component analysis and dynamic regression models. The results indicated reductions of up to 30% in the mean climatological streamflow patterns; and reductions in rainfall amounts of 22.9%, 13.3% and 12.4% at latitudes 9°, 12°and 13°South, respectively. Decreasing trends were found in simulations of hydropower generation under the influence of different climate variables. Thus, the hydroelectric system operates in contingency, due to the growing energy load demand resulting in more energy imports in Northeast Brazil.

TrajeR, an R package for cluster analysis of time series

trajeR, an R package for cluster analysis of time series

RNfuzzyApp: an R shiny RNA-seq data analysis app for visualisation, differential expression analysis, time-series clustering and enrichment analysis

RNA sequencing (RNA-seq) is a widely adopted affordable method for large scale gene expression profiling. However, user-friendly and versatile tools for wet-lab biologists to analyse RNA-seq data beyond standard analyses such as differential expression, are rare. Especially, the analysis of time-series data is difficult for wet-lab biologists lacking advanced computational training. Furthermore, most meta-analysis tools are tailored for model organisms and not easily adaptable to other species. With RNfuzzyApp, we provide a user-friendly, web-based R shiny app for differential expression analysis, as well as time-series analysis of RNA-seq data. RNfuzzyApp offers several methods for normalization and differential expression analysis of RNA-seq data, providing easy-to-use toolboxes, interactive plots and downloadable results. For time-series analysis, RNfuzzyApp presents the first web-based, fully automated pipeline for soft clustering with the Mfuzz R package, including methods to aid in cluster number selection, cluster overlap analysis, Mfuzz loop computations, as well as cluster enrichments. RNfuzzyApp is an intuitive, easy to use and interactive R shiny app for RNA-seq differential expression and time-series analysis, offering a rich selection of interactive plots, providing a quick overview of raw data and generating rapid analysis results. Furthermore, its assignment of orthologs, enrichment analysis, as well as ID conversion functions are accessible to non-model organisms.

RNfuzzyApp: an R shiny RNA-seq data analysis app for visualisation, differential expression analysis, time-series clustering and enrichment analysis

RNA sequencing (RNA-seq) is a widely adopted affordable method for large scale gene expression profiling. However, user-friendly and versatile tools for wet-lab biologists to analyse RNA-seq data beyond standard analyses such as differential expression, are rare. Especially, the analysis of time-series data is difficult for wet-lab biologists lacking advanced computational training. Furthermore, most meta-analysis tools are tailored for model organisms and not easily adaptable to other species. With RNfuzzyApp, we provide a user-friendly, web-based R shiny app for differential expression analysis, as well as time-series analysis of RNA-seq data. RNfuzzyApp offers several methods for normalization and differential expression analysis of RNA-seq data, providing easy-to-use toolboxes, interactive plots and downloadable results. For time-series analysis, RNfuzzyApp presents the first web-based, fully automated pipeline for soft clustering with the Mfuzz R package, including methods to aid in cluster number selection, cluster overlap analysis, Mfuzz loop computations, as well as cluster enrichments. RNfuzzyApp is an intuitive, easy to use and interactive R shiny app for RNA-seq differential expression and time-series analysis, offering a rich selection of interactive plots, providing a quick overview of raw data and generating rapid analysis results. Furthermore, its assignment of orthologs, enrichment analysis, as well as ID conversion functions are accessible to non-model organisms.