Clustering Large Applications Research Articles

Mapping of Village Population Profile with Schistosomiasis Cases Using Clustering Large Applications

Schistosomiasis is a tropical disease caused by Schistosoma mansoni (intestinal schistosomiasis) and Schistosoma haematobium (urogenital schistosomiasis). Schistosomiasis in Indonesia is endemic to Central Sulawesi and is commonly found in the Napu Valley and Bada Valley areas, which are administratively included in Poso District and Sigi District. One approach to obtain information on schistosomiasis endemic areas is by mapping the population profile of villages with schistosomiasis cases. This mapping is intended to provide an overview of the social and demographic conditions of villages with schistosomiasis cases. One of the many analysis methods that can be used is cluster analysis. Cluster analysis is a method for grouping data based on the extent of their similarities. Data with similar characteristics will be grouped together, while data with different characteristics will be placed in different groups. Among several types of methods in cluster analysis is Clustering Large Application (CLARA). CLARA is a clustering method which is more robust to unusual data and can be applied to handle large volumes of data. The results of this study are obtained two optimum clusters, each possessing distinct characteristics as determined by Schistosomiasis cases indicators. Cluster 1 with low schistosomiasis cases and cluster 2 with high schistosomiasis cases.

Unraveling determinants of severe anaphylaxis — A cluster analysis from a large national hospitalization database

BackgroundUnderstanding potential risk factors and phenotypes at-risk of anaphylaxis can tailor public health actions to prevent avoidable deaths. Therefore, our aim was to explore patterns of hospital admissions diagnosed with anaphylaxis in France from 2012 to 2021, across different age groups, using the national French hospital database and to evaluate potential determinants to propose phenotypes at-risk, with a particular view to severe cases. MethodsThis is a descriptive study, using data from the French Program for the Medicalization of Information Systems, from 2012 to 2021. We included all patients hospitalized for anaphylaxis using International Classification of Diseases (ICD)-10 codes listed as a primary diagnosis. Potential risk factors were selected according to corresponding ICD codes, available as secondary data of hospitalization. Severe anaphylaxis (SA) was defined based on the need of admissions in intensive care units. To explore the patterns and characteristics, we employed cluster analysis using the clustering large applications algorithm. Charlson comorbidity index (CCI) was used to evaluate comorbidities. ResultsDuring the 9-year study period, 78,453 patients had 86,654 admissions due to anaphylaxis for all-causes. Most of the cases were non-severe anaphylaxis (93.7%), 5.9% were SA, and 0.4% evolved to fatal anaphylaxis (FA). Rate of admissions for SA was 0.08/100,000/year. Seasonal trend of cases of anaphylaxis due to Hymenoptera venoms were shown. Increased severity and risk of death were significantly higher in the elderly. Food-induced anaphylaxis is more frequent in childhood, but SA occurred at adolescence and early adulthood. The CCI was higher in SA and FA. By applying the clustering algorithm CLARA, we identified 3 clusters distributed according to the severity. ConclusionMore than providing a snap shot of potential risk factors for anaphylaxis in France, we were able to suggest phenotypes of anaphylaxis of different severities and to propose actions based on our findings. Although further research is needed in the field, we strongly believe that the data presented may contribute to public health advancements to ensure quality of care of patients suffering from anaphylaxis.

EV charging load profile identification and seasonal difference analysis via charging sessions data of charging stations

The popularity of electric vehicles (EVs) brings environmental benefits, but their hard-to-estimate stochastic charging behaviors places additional diversity on grid load management. This paper proposes a procedure to identify typical charging load profiles (CLPs) via large scale charging session data from charging stations (CSs). The daily CLPs are computed from charging sessions, and a comprehensive similarity metric based on the weighting of Euclidean and Pearson correlation coefficient is proposed to achieve better clustering. The clustering is performed using the Clustering LARge Applications (CLARA) algorithm to accommodate large sample scenarios. Subsequently, hierarchical clustering of CSs is performed based on possible CLPs，and their CLPs are estimated by Monte Carlo simulation. The performance of the proposed method is tested and evaluated with over 340,000 charging sessions from 109 CSs in Wuhan at central China, and seasonal differences in CLPs are explored. The results show that the method of mining typical CLPs from charging sessions is effective, 17 typical CLPs are identified in different seasons, which provide effective information on the fluctuation and magnitude of daily power demand, the charging power demand also shows significant seasonal differences, and good accuracy is achieved by dividing the CSs into different groups for load estimation.

Clustering countries according to the logistics performance index

This study aims to cluster countries according to the logistics performance index (LPI) 2018 data. LPI is one of the indicators for measuring logistics sector performance based on a survey developed by the World Bank since 2007 and has been widely accepted throughout the world. There are 160 countries involved in organizing the 2018 edition of the LPI. It helps countries to understand their current position and develop strategies and policies to improve their performance in world trade. Three clustering algorithms (i.e., k-means, k-medoids, and clustering large applications) are used. To obtain the optimal number of clusters, the elbow method is used. According to the elbow method, the optimal cluster is three for those three clustering algorithms. The countries belong to the first cluster are considered the best performers, while countries in the third cluster are the worst performance in terms of logistics performance. This study is expected to give an insight into how to implement clustering algorithms into the real-world data set and how to interpret the results

Clustering objects for spatial data mining: a comparative study

Spatial data mining (SDM) is searching important relationships and characteristics that can clearly exist in spatial databases. This content aims to compare object clustering algorithms for spatial data mining, before identifying the most efficient algorithm. To this end, this paper compare k-means, Partionning Around Medoids (PAM) and Clustering Large Applications based on RANdomized Search (CLARANS) algorithms based on computing time. Experimental results indicate that, CLARANS is very efficient and effective.

Unsupervised Machine Learning with Cluster Analysis in Patients Discharged after an Acute Coronary Syndrome: Insights from a 23,270-Patient Study

Characterization and management of patients admitted for acute coronary syndromes (ACS) remain challenging, and it is unclear whether currently available clinical and procedural features can suffice to inform adequate decision making. We aimed to explore the presence of specific subsets among patients with ACS. The details on patients discharged after ACS were obtained by querying an extensive multicenter registry and detailing patient features, as well as management details. The clinical outcomes included fatal and nonfatal cardiovascular events at 1-year follow-up. After missing data imputation, 2 unsupervised machine learning approaches (k-means and Clustering Large Applications [CLARA]) were used to generate separate clusters with different features. Bivariate- and multivariable-adjusted analyses were performed to compare the different clusters for clinical outcomes. A total of 23,270 patients were included, with 12,930 cases (56%) of ST-elevation myocardial infarction (STEMI). K-means clustering identified 2 main clusters: a first 1 including 21,998 patients (95%) and a second 1 including 1,282 subjects (5%), with equal distribution for STEMI. CLARA generated 2 main clusters: a first 1 including 11,268 patients (48%) and a second 1 with 12,002 subjects (52%). Notably, the STEMI distribution was significantly different in the CLARA-generated clusters. The clinical outcomes were significantly different across clusters, irrespective of the originating algorithm, including death reinfarction and major bleeding, as well as their composite. In conclusion, unsupervised machine learning can be leveraged to explore the patterns in ACS, potentially highlighting specific patient subsets to improve risk stratification and management.

Study on facility location of air taxi skyports using a prescriptive analytics approach

The current study identifies infrastructure locations for air taxi operations in New York City (NYC) using a prescriptive analytics approach. Based on the potential air taxi demand estimated by prior studies, 14 unique sites are identified by an unsupervised machine learning technique called clustering large applications (CLARA) in Step-1. However, developing all the suggested stations simultaneously might be challenging for any business, and therefore, an integer programming model is developed in Step-2 to propose opening these centers in multiple phases, considering parameters such as rental cost, daily number of trips, easy access to road facilities, and employee salary, while maximizing the demand satisfaction. Sensitivity analysis is then conducted to identify the impact of various market penetration strategies (such as aggressive, defensive and balanced), inflation in the rental cost and employee salary, and the emergence of latent demand on the phased opening of the facilities. The clustering algorithm results indicate seven air taxi facilities to be established in the Manhattan borough, four in Queens, two in Brooklyn, and one in New Jersey, spread across three phases. Larger vertiports are suggested for airports, such as John F Kennedy and LaGuardia international airports, since they serve approximately 55% of the overall demand. Results also showed that varying percentages of the budgeted rental cost and employee salary have minimal effect on the total number of stations.

Indonesian Territory Clustering Based On Harvested Area and Rice Productivity Using Clustering Algorithm

Rice (Latin: Oryza sativa) is one of the most important cultivated plants in civilization. This plant is the main commodity for almost all Indonesian people. Indonesia is in third place as the largest rice producing country in the world. However, based on data from the Statistics Indonesia, Indonesia will still import rice until 2022. The transfer of paddy fields is one of the reasons why Indonesia is still importing rice to this day. Many lands that used to be paddy fields have turned into airports, industrial land, housing, and so on. Rice production is one of the important topics to be discussed in order to develop rice production in areas that are still relatively low. The purpose of this research is to classify cities/regencies in Indonesia based on rice production data in 2021. In this study, three clustering methods were used, namely, Partitioning Around Medoid (PAM), Clustering Large Applications (CLARA) and Fuzzy C-Means (FCM). Then the three methods are compared based on their silhouette coefficient values. The best obtained method is FCM method with two clusters and a silhouette value of 0.828. Results clustering with the best method is used as a reference in making maps clustering. Areas that are still relatively low are expected to increase rice productivity.

Evaluating Partitioning Based Clustering Methods for Extended Non-negative Matrix Factorization (NMF)

Data is humongous today because of the extensive use of World Wide Web, Social Media and Intelligent Systems. This data can be very important and useful if it is harnessed carefully and correctly. Useful information can be extracted from this massive data using the Data Mining process. The information extracted can be used to make vital decisions in various industries. Clustering is a very popular Data Mining method which divides the data points into different groups such that all similar data points form a part of the same group. Clustering methods are of various types. Many parameters and indexes exist for the evaluation and comparison of these methods. In this paper, we have compared partitioning based methods K-Means, Fuzzy C-Means (FCM), Partitioning Around Medoids (PAM) and Clustering Large Application (CLARA) on secure perturbed data. Comparison and identification has been done for the method which performs better for analyzing the data perturbed using Extended NMF on the basis of the values of various indexes like Dunn Index, Silhouette Index, Xie-Beni Index and Davies-Bouldin Index.

56 ANALYSIS OF A 23,270-PATIENT STUDY: UNSUPERVISED MACHINE LEARNING WITH CLUSTER ANALYSIS IN PATIENTS DISCHARGED AFTER AN ACUTE CORONARY SYNDROME

Abstract Background Characterization and management of patients with acute coronary syndromes (ACS) remain challenging, and it is unclear whether currently available clinical and procedural features can suffice to inform adequate decision making. Methods Details on patients discharged after an ACS were obtained by querying an extensive multicenter registry, detailing patient features as well as management details. Clinical outcomes included fatal and non-fatal cardiovascular events at 1-year follow-up. After missing data imputation, two unsupervised machine learning approaches (k-means and Clustering Large Applications [CLARA]) were used to generate separate clusters with different features. Bivariate and multivariable-adjusted analyses were performed to compare different clusters for clinical outcomes. Findings: 23,270 patients discharged after ACS were included. Two clusters were identified by k-mean algorithm (k1 and k2), and two clusters by CLARA algorithm (C1 and C2). Differences in 2-years outcomes between k1 and k2 and between C1 and C2 were substantial. K2 cluster (N=21,988) in comparison with k1 cluster (N=1,282) had significantly higher occurrence of death (9.5% vs 3.8%, p<0.001), reinfarction (7.2% vs 3.7%, p<0.001), and major bleeding (6.0% vs 3.0%, p<0.001). Similarly, C1 cluster (N=11,268) showed a worse prognosis than C2 cluster (N=12,002): death (4.8% vs 3.5%, p<0.001), reinfarction (4.5% vs 3.4%, p<0.001), and major bleeding (3.6% vs 2.8%, p=0.001). Most associations did not hold at multivariable analysis based on supervised learning techniques, with the exception of major bleeding (odds ratio=1.37 [95% confidence interval 1.02-1.83] for the k1/C1 subcluster vs k1/C2 subcluster, p=0.039. Conclusions A machine-learning based clustering approach is effective at face value to inform on the prognosis of patients with ACS managed invasively. These findings can be leveraged to inform decision-making in this setting, but also highlight the potential role of cluster analysis in first-in-man, and preapproval studies of medical devices.

Clinical-epidemiologic variation in patients treated in the first and second wave of COVID-19 in Lambayeque, Peru: A cluster analysis

ObjectivesTo identify differences in the clinical and epidemiologic characteristics of patients during the first and second waves of the COVID-19 pandemic at the EsSalud Lambayeque health care network, Peru. MethodsAn analytical cross-sectional study of 53,912 patients enrolled during the first and second waves of COVID-19 was conducted. Cluster analysis based on clustering large applications (CLARA) was applied to clinical-epidemiologic data presented at the time of care. The two pandemic waves were compared using clinical-epidemiologic data from epidemiologic surveillance. ResultsCluster analysis identified four COVID-19 groups with a characteristic pattern. Cluster 1 included the largest number of participants in both waves, and the participants were predominantly female. Cluster 2 included patients with gastrointestinal, respiratory, and systemic symptoms. Cluster 3 was the “severe” cluster, characterized by older adults and patients with dyspnea or comorbidities (cardiovascular, diabetes, obesity). Cluster 4 included asymptomatic, pregnant, and less severe patients. We found differences in all clinical-epidemiologic characteristics according to the cluster to which they belonged. ConclusionUsing cluster analysis, we identified characteristic patterns in each group. Respiratory, gastrointestinal, dyspnea, anosmia, and ageusia symptoms were higher in the second COVID-19 wave than the first COVID-19 wave.

Clustering compositional data using Dirichlet mixture model.

A model-based clustering method for compositional data is explored in this article. Most methods for compositional data analysis require some kind of transformation. The proposed method builds a mixture model using Dirichlet distribution which works with the unit sum constraint. The mixture model uses a hard EM algorithm with some modification to overcome the problem of fast convergence with empty clusters. This work includes a rigorous simulation study to evaluate the performance of the proposed method over varied dimensions, number of clusters, and overlap. The performance of the model is also compared with other popular clustering algorithms often used for compositional data analysis (e.g. KMeans, Gaussian mixture model (GMM) Gaussian Mixture Model with Hard EM (Hard GMM), partition around medoids (PAM), Clustering Large Applications based on Randomized Search (CLARANS), Density-Based Spatial Clustering of Applications with Noise (DBSCAN) etc.) for simulated data as well as two real data problems coming from the business and marketing domain and physical science domain, respectively. The study has shown promising results exploiting different distributional patterns of compositional data.

Clinical characteristics with inflammation profiling of long COVID and association with 1-year recovery following hospitalisation in the UK: a prospective observational study

BackgroundNo effective pharmacological or non-pharmacological interventions exist for patients with long COVID. We aimed to describe recovery 1 year after hospital discharge for COVID-19, identify factors associated with patient-perceived recovery, and identify potential therapeutic targets by describing the underlying inflammatory profiles of the previously described recovery clusters at 5 months after hospital discharge.MethodsThe Post-hospitalisation COVID-19 study (PHOSP-COVID) is a prospective, longitudinal cohort study recruiting adults (aged ≥18 years) discharged from hospital with COVID-19 across the UK. Recovery was assessed using patient-reported outcome measures, physical performance, and organ function at 5 months and 1 year after hospital discharge, and stratified by both patient-perceived recovery and recovery cluster. Hierarchical logistic regression modelling was performed for patient-perceived recovery at 1 year. Cluster analysis was done using the clustering large applications k-medoids approach using clinical outcomes at 5 months. Inflammatory protein profiling was analysed from plasma at the 5-month visit. This study is registered on the ISRCTN Registry, ISRCTN10980107, and recruitment is ongoing.Findings2320 participants discharged from hospital between March 7, 2020, and April 18, 2021, were assessed at 5 months after discharge and 807 (32·7%) participants completed both the 5-month and 1-year visits. 279 (35·6%) of these 807 patients were women and 505 (64·4%) were men, with a mean age of 58·7 (SD 12·5) years, and 224 (27·8%) had received invasive mechanical ventilation (WHO class 7–9). The proportion of patients reporting full recovery was unchanged between 5 months (501 [25·5%] of 1965) and 1 year (232 [28·9%] of 804). Factors associated with being less likely to report full recovery at 1 year were female sex (odds ratio 0·68 [95% CI 0·46–0·99]), obesity (0·50 [0·34–0·74]) and invasive mechanical ventilation (0·42 [0·23–0·76]). Cluster analysis (n=1636) corroborated the previously reported four clusters: very severe, severe, moderate with cognitive impairment, and mild, relating to the severity of physical health, mental health, and cognitive impairment at 5 months. We found increased inflammatory mediators of tissue damage and repair in both the very severe and the moderate with cognitive impairment clusters compared with the mild cluster, including IL-6 concentration, which was increased in both comparisons (n=626 participants). We found a substantial deficit in median EQ-5D-5L utility index from before COVID-19 (retrospective assessment; 0·88 [IQR 0·74–1·00]), at 5 months (0·74 [0·64–0·88]) to 1 year (0·75 [0·62–0·88]), with minimal improvements across all outcome measures at 1 year after discharge in the whole cohort and within each of the four clusters.InterpretationThe sequelae of a hospital admission with COVID-19 were substantial 1 year after discharge across a range of health domains, with the minority in our cohort feeling fully recovered. Patient-perceived health-related quality of life was reduced at 1 year compared with before hospital admission. Systematic inflammation and obesity are potential treatable traits that warrant further investigation in clinical trials.FundingUK Research and Innovation and National Institute for Health Research.

The impact of the COVID-19 pandemic on O-D flow and airport networks in the origin country and in Northeast Asia

The ongoing COVID-19 pandemic has posed a global threat to human health. In order to prevent the spread of this virus, many countries have imposed travel restrictions. This difficult situation has dramatically affected the airline industry by reducing the passenger volume, number of flights, airline flow patterns, and even has changed the entire airport network, especially in Northeast Asia (because it includes the original disease seed). However, although most scholars have used conventional statistical analysis to describe the changes in passenger volume before and during the COVID-19 outbreak, very few of them have applied statistical assessment or time series analysis, and have not even examined how the impact may be different from place to place. Therefore, the purpose of this study was to identify the impact of COVID-19 on the airline industry and affected areas (including the origin-destination flow and the airport network). First, a Clustering Large Applications (CLARA) algorithm was used to group numerous origin-destination (O-D) flow patterns based on their characteristics and to determine if these characteristics have changed the severity of the impact of each cluster during the COVID-19 outbreak. Second, two statistical tests (the paired t-test and the Wilcoxon signed-rank test) were utilized to determine if the entire airport network and the top 30 hub airports changed during COVID-19. Four centrality measurement indices (degree, closeness, eigenvector, and betweenness centrality) of the airports were used to assess the entire network and ranking of individual hub airports. The study data, provided by The Official Aviation Guide (OAG) from December 2019 to April 2020, indicated that during the COVID-19 outbreak, there was a decrease in passenger volume (60%–98.4%) as well as the number of flights (1.5%–82.6%). However, there were no such significant changes regarding the popularity ranking of most airports during the outbreak. Before this occurred (December 2019), most hub airports were in China (April 2020), and this trend remain similar during the COVID-19 outbreak. However, the values of the centrality measurement decreased significantly for most hub airports due to travel restrictions issued by the government.

Incorporating Spatial and Temporal Correlations to Improve Aggregation of Decentralized Day-Ahead Wind Power Forecasts

In some electricity markets, individual wind farms are obliged to provide point forecasts to the power purchaser or system operator. These decentralized forecasts are usually based on on-site meteorological forecasts and measurements, and thus optimized for local conditions. Simply adding decentralized forecasts may not capture some of the spatial and temporal correlations of wind power, thereby lowering the potential accuracy of the aggregated forecast. This paper proposes the explanatory variables that are used to train the kernel density estimator and conditional kernel density estimator models to derive day-ahead aggregated point and probabilistic wind power forecasts from decentralized point forecasts of geographically distributed wind farms. The proposed explanatory variables include (a) decentralized point forecasts clustered using the clustering large applications algorithm to reduce the high-dimensional matrices, (b) hour of day and month of year to account for diurnal and seasonal cycles, respectively, and (c) atmospheric states derived from self-organizing maps to represent large-scale synoptic circulation climatology for a study area. The proposed methodology is tested using the day-ahead point forecast data obtained from 29 wind farms in South Africa. The results from the proposed methodology show a significant improvement as compared to simply adding the decentralized point forecasts. The derived predictive densities are shown to be non-Gaussian and time-varying, as expected given the time-varying nature of wind uncertainty. The proposed methodology provides system operators with a method of not only producing more accurate aggregated forecasts from decentralized forecasts, but also improving operational decisions such as dynamic operating reserve allocation and stochastic unit commitment.

Physical, cognitive, and mental health impacts of COVID-19 after hospitalisation (PHOSP-COVID): a UK multicentre, prospective cohort study

SummaryBackgroundThe impact of COVID-19 on physical and mental health and employment after hospitalisation with acute disease is not well understood. The aim of this study was to determine the effects of COVID-19-related hospitalisation on health and employment, to identify factors associated with recovery, and to describe recovery phenotypes.MethodsThe Post-hospitalisation COVID-19 study (PHOSP-COVID) is a multicentre, long-term follow-up study of adults (aged ≥18 years) discharged from hospital in the UK with a clinical diagnosis of COVID-19, involving an assessment between 2 and 7 months after discharge, including detailed recording of symptoms, and physiological and biochemical testing. Multivariable logistic regression was done for the primary outcome of patient-perceived recovery, with age, sex, ethnicity, body-mass index, comorbidities, and severity of acute illness as covariates. A post-hoc cluster analysis of outcomes for breathlessness, fatigue, mental health, cognitive impairment, and physical performance was done using the clustering large applications k-medoids approach. The study is registered on the ISRCTN Registry (ISRCTN10980107).FindingsWe report findings for 1077 patients discharged from hospital between March 5 and Nov 30, 2020, who underwent assessment at a median of 5·9 months (IQR 4·9–6·5) after discharge. Participants had a mean age of 58 years (SD 13); 384 (36%) were female, 710 (69%) were of white ethnicity, 288 (27%) had received mechanical ventilation, and 540 (50%) had at least two comorbidities. At follow-up, only 239 (29%) of 830 participants felt fully recovered, 158 (20%) of 806 had a new disability (assessed by the Washington Group Short Set on Functioning), and 124 (19%) of 641 experienced a health-related change in occupation. Factors associated with not recovering were female sex, middle age (40–59 years), two or more comorbidities, and more severe acute illness. The magnitude of the persistent health burden was substantial but only weakly associated with the severity of acute illness. Four clusters were identified with different severities of mental and physical health impairment (n=767): very severe (131 patients, 17%), severe (159, 21%), moderate along with cognitive impairment (127, 17%), and mild (350, 46%). Of the outcomes used in the cluster analysis, all were closely related except for cognitive impairment. Three (3%) of 113 patients in the very severe cluster, nine (7%) of 129 in the severe cluster, 36 (36%) of 99 in the moderate cluster, and 114 (43%) of 267 in the mild cluster reported feeling fully recovered. Persistently elevated serum C-reactive protein was positively associated with cluster severity.InterpretationWe identified factors related to not recovering after hospital admission with COVID-19 at 6 months after discharge (eg, female sex, middle age, two or more comorbidities, and more acute severe illness), and four different recovery phenotypes. The severity of physical and mental health impairments were closely related, whereas cognitive health impairments were independent. In clinical care, a proactive approach is needed across the acute severity spectrum, with interdisciplinary working, wide access to COVID-19 holistic clinical services, and the potential to stratify care.FundingUK Research and Innovation and National Institute for Health Research.

Retrospective identification of latent subgroups of emergency department patients: A machine learning approach.

This research aims to (i) identify latent subgroups of ED presentations in Australian public EDs using a data-driven approach and (ii) compare clinical, socio-demographic and time-related characteristics of ED presentations broadly using the subgroups. We examined presentations to four public hospital EDs in Queensland from 2009 to 2014. An unsupervised machine learning algorithm, Clustering Large Applications, was used to cluster ED presentations. There were six subgroups common across the EDs, primarily distinguishable by age, and subsequently by triage category, ED length of stay, arrival mode, departure status and several time-related attributes. Around 10% to 30% of the total presentations had high resource utilisation, with half of these from older patients (55+ years). ED resource utilisation per population was highest among the oldest cohort (75+ years). Children and young adults more frequently presented to the ED outside general-practitioner hours, mostly on Sundays. Older persons were more likely to present at any time, rather than specific hours, days or seasons. ED service performance measured against commonly used access-target indicators were rarely satisfied for older people and frequently satisfied for children. Clustering Large Applications is effective in finding latent groups in large-scale mixed-type data, as demonstrated in the present study. Six types of ED presentations were identified and described using clinically relevant characteristics. The present study provides evidence for policy makers in Australia to develop alternative ED models of care tailored around the care needs of the differing groups of patients and thereby supports the sustainable delivery of acute healthcare.

Solar farm voltage anomaly detection using high-resolution [formula omitted]PMU data-driven unsupervised machine learning

The usual means of solar farm condition monitoring are limited by the typically poor quality and low-resolution data collected. A micro-synchrophasor measurement unit has been adapted and integrated with a power quality monitor to provide the high-resolution, high-precision, synchronised time-series data required by analysts to significantly improve solar farm performance and to better understand their impact on distribution grid behaviour. Improved renewable energy generation at large solar photovoltaic facilities can be realised by processing the enormous amounts of high-quality data using machine learning methods for automatic fault detection, situational awareness, event forecasting, operational tuning, and planning condition-based maintenance. The limited availability of existent data knowledge in this sector and legacy performance issues steered our exploration of machine learning based approaches to the unsupervised direction. A novel application of the Clustering Large Applications (CLARA) algorithm was employed to categorise events from the large datasets collected. CLARA has been adapted to recognise solar site specific behaviour patterns, abnormal voltage dip and spike events using the multiple data streams collected at two utility-scale solar power generation sites in England. Fourteen days of empirical field data (seven consecutive summer days plus seven consecutive winter days) enabled this analytical research and development approach. Altogether, ∼725 million voltage measurement data points were investigated, and automatic voltage anomaly detection demonstrated.

Numerical soil horizon classification from South Africa’s legacy database

Allocation of soil profiles to one or another soil class depends on the surveyor’s experience and knowledge. Inconsistent class designations can affect the downstream uses of soil information such as hydrological and ecological applications. Therefore, numerical classifications can be useful in creating property-based soil clusters. Quantitative classification of soil horizons can also be beneficial for guiding physicochemical thresholds used in taxonomic criteria. This study aimed at taking a step in applying numerical classification techniques to the South African soil legacy database to quantitatively cluster horizons. For each master horizon, the database was clustered into new “diagnostic” horizons through Partitioning Around Medoids (PAM) implemented in the Clustering Large Applications algorithm (CLARA). This algorithm has the benefit of using multiple subsets of the data making a more reliable initiation of medoids and it is a non-parametric approach making it more robust to outliers. To determine the optimal number of clusters, the silhouette width for 1 through 11 clusters was calculated. The number of clusters with the largest silhouette width was taken as the optimal number of clusters. The results of this study show that for O, E, and G (gleyed) horizons, the algorithm performs relatively well with the first two principle components capturing almost half of the variation in the data. However, for the A, B, and C horizons, the algorithm struggled to separate clusters sufficiently. The A horizon clusters only broadly corresponded to environmental factors (geography, climate, elevation, and geology) and the large clusters straddled numerous taxonomic thresholds. Moreover, the B horizon clusters correspond relatively well with environmental factors. Limitations identified, include bias in geographic location of modal soil profiles of the database, bias in the South African classification system, how to handle non-globular clusters, and outliers which add leverage to the clusters.

CLUSTERING OF EARTHQUAKE RISK IN INDONESIA USING K-MEDOIDS AND K-MEANS ALGORITHMS

Earthquake is the shaking of the earth's surface due to the shift in the earth's plates. This disaster often happens in Indonesia due to the location of the country on the three largest plates in the world and nine small others which meet at an area to form a complex plate arrangement. An earthquake has several impacts which depend on the magnitude and depth. This research was, therefore, conducted to classify earthquake data in Indonesia based on the magnitudes and depths using one of the data mining techniques which is known as clustering through the application of k-medoids and k-means algorithms. However, k-medoids group data into clusters with medoid as the centroid and it involves using clustering large application (CLARA) algorithm while k-means divide data into k clusters where each object belongs to the cluster with the closest average. The results showed the best clustering for earthquake data in Indonesia based on magnitude and depth is the CLARA algorithm and five clusters were found to have total members of 2231, 1359, 914, 2392, and 199 objects for cluster 1 to cluster 5 respectively.