Search

Water resource management is one of the biggest challenges that are being faced, such as a warming climate, arid land, and toxic chemicals in the water. It is essential to deal with water resource management urgently. In this article, researchers mainly focus on monitoring the water quality in the Lam Pa Thao dam, Chaiyaphum, Thailand. The farmer in that area directly affected by the water quality in the dam because they raise fish in floating fish cages. To prevent losses from fish farming, they should have the ability to monitor and control the factors that affect the water quality. As a result, the farmer can monitor the water quality and the monitor system can report to the farmer in time. In this case, to monitor the water quality, researchers designed the buoys, which is the internet of things device, to collect data from the Lam Pa Thao dam. researchers collected the water quality data from January - March 2021, including 13,608 instances. The five important parameters were obtained, including dissolved oxygen, temperature, pH, total dissolved solids, and electric conductivity. Due to the number of parameters, researchers decided not to apply dimension reduction. In these experiments, researchers proposed using K-means clustering algorithms to group the water data into appropriate clusters. For the K-Means algorithm, we calculated the silhouette coefficient to analyze the effectiveness of cluster separation. The best cluster that was grouped using the K-means algorithm achieved the silhouette score of 0.6839. Furthermore, researchers evaluated the K-means algorithm on Charles river and Fitzroy river datasets. It obtained the silhouette score of 0.5489 and 0.6589, respectively.

The coronavirus COVID-19 has been affected all the countries and territories in 2020. In this study we cluster European countries according to the cumulative relative number of European COVID-19 patients. The clustering is based on publicly available data published at European Centre for Disease Prevention and Control website and performed by three clustering methods: K-means, agglomerative and BIRCH clustering. Clustering performance, evaluated by Silhouette Coefficient value, shows satisfying accuracy of the obtained clusters. The results presented in this study can be useful to public health officers and practitioners to easier deal with COVID-19 challenges.

This paper mainly analyzes the image data of multiple lesions. Analyses were conducted on the raw data such as the frequency statistics of lesions type, the distribution of lesion size, and the age distribution of patients. The analysis data mainly include lesion diameters pixel, coarse lesion type, patient age. Based on this foundation, silhouette coefficient and sum of the squared errors (SSE) are adopted to analysis the optimal K value for anchor boxes cluster. Through the analysis of medical image data, it can provide the foundation for predicting the growth of lesions. It can also be used to classify the lesions, and guide the development of general lesion detector based on database in the next work. This work can help doctors to make clinical diagnosis more comprehensively and improve the efficiency of clinical diagnosis.

Traditional fuzzy C-means (FCM) clustering algorithm is sensitive to initial clustering center, and the number of clusters need to be set artificially in advance. For these reasons, we propose an improved FCM algorithm (AMMF) that can determine the number of clusters automatically. Firstly, the proposed algorithm uses the affinity propagation clustering algorithm to obtain coarse number of clusters, which are taken as the upper limit of searching the best number of clusters. Secondly, by the improved maximum and minimum distance algorithm obtains some representative sample points as the initial clustering centers of the FCM algorithm. Lastly, we use Silhouette Coefficient to analyze the quality of clustering to determine the optimal number of clusters automatically. Experimental results show that the AMMF algorithm has significantly better clustering performance than other improved FCM based algorithms, and improves the stability of the clustering results.

Nowadays, an industry’s expectation that’s often not fulfilled by job applicants require companies to actively cooperate with universities, one of the reasons is through employees that is considered to have good performance to find talents within their alma mater. This research aims to analyze job vacancy information uploaded by graduates for juniors in their university that can be mapped into a graduate’s profile and evaluation materials in making a curriculum. Collected job vacancy information from several communication media are generally unstructured data which requires it to be preprocessed first through a data mining convention to produce several terms ready to be processed, continued with implementation of TF-IDF, feature extraction using PCA, and grouping using k-Means algorithm. The clustering analysis found 3 job clusters i.e. developer, teacher and researcher/lecturer as job vacancies that frequently shared by graduates. This result obtained from clustering analysis using 10 words as a minimum document frequency based on Elbow Method and Silhouette Coefficient analysis.

the quality of students' performance clusters relates to the accuracy of students being in groups based on their performance. However, the resulting quality sometimes needs to be improved because the clustering process involves features that are not dominant. Furthermore, in the previous works, measurement of the quality of the clusters in unsupervised evaluation often only uses one measure. Therefore, this paper focuses to enhance the quality of clusters by eliminating features that are irrelevant by applying the feature selection method called the Gini Index. Meanwhile, in this paper, the clustering method applied is K-means for the mining process. Then, we propose the evaluation process measured by three metrics, namely: silhouette coefficient, ANOVA, and t-test. The experimental results show that the Gini Index can improve the quality of clusters based on the three proposed metrics.

In order to analyze the relationship between the national epidemic strategy response to COVID-19 and the severity of the current epidemic, this paper firstly collects the national strategic data responds to the COVID-19 epidemic situation and the resulting data of COVID-19 in these nations, then use StandardScaler normalization method to pre-process the data, and use K-Means Clustering, Agglomerative Clustering and Density-Based Spatial Clustering of Applications with Noise (DBSCAN) clustering algorithm to perform cluster analysis on the above two types of data, use Silhouette Coefficient index, Calinski Harabasz Score, Davies bouldin score to evaluate the clustering results, and take the clustering results with the best clustering effect to analyze the impact of national epidemic policies on the current epidemic results, finally use cases to prove the effectiveness and feasibility of the model, and provide data support and strategic support for the prevention and control of the epidemic similar to COVID-19 in the future.

The classification of transformer districts is very important for line loss evaluation, load characteristic analysis and so on. In this paper, a Clustering technique of K-Means++ based on t-SNE dimension reduction is proposed and realized by programming. Firstly, the original matrix representing the electrical characteristic index of the transformer district are established. Secondly, the original normalized matrix is reduced by t-SNE dimension reduction method to obtain the main characteristic matrix. After that, K-Means++ is used to cluster the main characteristic matrix. Then, the accuracy of clustering results is evaluated by Silhouette Coefficient (SC) and Calinski-Harabaz (CH) index. Finally, selecting 1959 transformer districts as an example, simulation and calculation are performed to verify the effectiveness and practicable of the K-Means++ algorithm based on t-SNE dimension reduction in transformer district clustering.

Hadith is the second source of Islamic religious law after the Al-Qur’an, in the hadith there are many chapters that discuss several cases and will be interesting to be combined with data mining techniques, especially text mining in order to group the hadith into several groups based on Matan (content hadith) automatically. Clustering is a technique of grouping data based on criteria, in clustering has several methods including K-Means and Fuzzy C-Means. This research will try to group the Indonesian translation of Hadith texts and compare K-Means and Fuzzy C-Means algorithms with some parameters and experiments that are determined. This comparison is used to determine the most accurate method in the Hadith clustering. The results of this research indicate that some of the parameters used to affect the results of cluster evaluation, especially in reducing data dimensions. In Silhouette Coefficient and F-Measure calculations, the Fuzzy C-Means method has an accuracy of 0.83079 and 0.97128 while the K-Means method has an accuracy of 0.67828 and 0.95078 with the results above show that the Fuzzy C-Means method is better in grouping the Indonesian hadith text.

New protein sequences and structures are made available on an accelerated pace in public databases, which demands automatic reliable methods to make sense of this large volume of data. Many strategies for protein structural classification were proposed in the last years, using descriptors based on sequence and structure, as physicochemical properties of residues, solvent accessibility, and contact maps to perceive patterns that are able to characterize and predict protein structural families. In this paper, we assess whether it is possible to perform a structural classification of proteins through unsupervised learning using features successfully proposed in a well-established protein structural classifier. We conducted experiments using 5 clustering algorithms from different paradigms. The best result was achieved using agglomerative clustering with the Complete-Link algorithm (silhouette coefficient 0.843). Group labels were compared to protein superfamilies (CATH database) through the Fowlkes-Mallows Index to check if groups resulting from unsupervised learning correlated well with protein superfamilies. Our results show that even though successful, clustering analysis is not able to separate or classify the protein dataset in structural superfamilies in an efficient manner.

Forest, land, or residential fire is a familiar phenomenon in Indonesia for last decade. The high number of fire incidents in Indonesia requires attention from the government so that any natural disasters such as forest fires can be resolved. These fire incidents can be analyzed since the data has already been obtained and recorded from satellite. Unfortunately, the data is too large to be analyzed as it was. Based on data obtained from the EOSDIS website, recorded as many as 289,256 fire spots occur in the region of Sumatra in the timeframe between 2001 and 2014. It needs an algorithm to segment the data or clusters the data so that large data can be processed into good information for the user. In this study, a comparative study of clustering algorithms between the K-Means and the Isodata was conducted. Both algorithms used in this study were assessed based on the quality of the clusters produced, which is calculated using Silhouette Coefficient (SC). The final result value of Silhouette Coefficient the K-Means method is 0.999997187, and the Isodata method is 0.999957161. so in this case, K-Means algorithm has a higher SC value compared to the Isodata algorithm in clustering the data of fire spots with a small SC value difference.

In many organizations, enquiry emails from customers remain unanswered due to lack of patience and availability of a respondent. Generating FAQs from email repositories with lot of enquiry emails will be beneficial. However, manual generation of FAQs by experts is a time consuming and strenous job. Hence automatic generation of FAQs is a necessity. Automatic generation of FAQs require effective categorization of emails which is challenging since the emails are written by different people with heterogenous cognition levels. In this paper, we propose a framework using Non-negative Matrix Factorization (NMF) and k-means that groups emails into clusters which can be used for FAQ generation. The proposed framework determines not only the broad topic under which the emails have to be tagged but also categorizes the emails into clusters with similar sub contents. The number of clusters was determined by the elbow method whereas the number of topics was fixed by calculating the percentage of relevant topics. The average Silhouette coefficient score of the resulting clusters was found to be 0.52 indicating reasonably good clusters. Also, the Silhouette coefficient score of the proposed method increased by 36.82 % compared to k-means.

In this study, universities are clustered based on the quantity and title of the academics employed by Turkish Universities. The data comes from both state and private universities and vocational schools. For clustering, partitional and hierarchical clustering methods were used. The dataset was obtained from the YOK academic search website using “web scraping” techniques implemented by the authors. Agglomerative clustering technique was found to yield better results regarding cluster sizes and intracluster distances. The Silhouette coefficient is used for clustering quality measurement. As a result of the study, Turkish universities were mainly composed of four clusters. These are instructor dominant universities, just instructor based vocational schools, professor and associate professor dominated universities and research assistant dominated universities. The results of this study may be used for academic staff employment planning in Turkish Universities.

One of graduation requirements at university is completing undergraduate thesis. At Industrial Engineering Universitas Ahmad Dahlan, undergraduate thesis titles are documented by thesis coordinator. The problem is that students are less knowledgeable on thesis topics, so they do not really know the previous students’ thesis topics. Based on the problem, this research aims at developing a program to classify thesis title so the knowledge on the trend of thesis title topic can be got The method used in this research was K-Means clustering, while range measurement method used was cosine similarity. The testing used Silhouette Coefficient method. The phases from text mining were tokenizing, filtering, stemming, similarity, classifying, testing. The result of this research is a program that can process the title data into trend group pattern of thesis title topic.From 138 data obtained, there are three clusters arranged based on the field on Industrial Engineering study program. Silhouette Coefficient testing shows score of 0.5674 that shows the clustering result is classified low. It occurs since the textual data of the thesis title is too widely distributed, so the title has relatively low similarity score

Cluster analysis is an unsupervised learning method that constitutes a cornerstone of an intelligent data analysis process. Clustering categorical data is an important research area data mining. In this paper we propose a novel algorithm to cluster categorical data. Based on the minimum dissimilarity value objects are grouped into cluster. In the merging process, the objects are relocated using silhouette coefficient. Experimental results show that the proposed method is efficient.