• The Korean Journal of Applied Statistics
• /
• v.22 no.4
• /
• pp.723-733
• /
• 2009

The crucial problems of K-means clustering are deciding the number of clusters and initial centroids of clusters. Hence, the steps of K-means clustering are generally consisted of two-stage clustering procedure. The first stage is to run hierarchical clusters to obtain the number of clusters and cluster centroids and second stage is to run nonhierarchical K-means clustering using the results of first stage. Here we provide automated K-means clustering procedure to be useful to obtain initial centroids of clusters which can also be useful for large data sets, and provide software program implemented using R.

• The Korean Journal of Applied Statistics
• /
• v.34 no.6
• /
• pp.905-922
• /
• 2021

Cluster analysis is one of unsupervised learning techniques used for discovering clusters when there is no prior knowledge of group membership. K-means, one of the commonly used cluster analysis techniques, may fail when the number of variables becomes large. In such high-dimensional cases, it is common to perform tandem analysis, K-means cluster analysis after reducing the number of variables using dimension reduction methods. However, there is no guarantee that the reduced dimension reveals the cluster structure properly. Principal component analysis may mask the structure of clusters, especially when there are large variances for variables that are not related to cluster structure. To overcome this, techniques that perform dimension reduction and cluster analysis simultaneously have been suggested. This study proposes probabilistic reduced K-means, the transition of reduced K-means (De Soete and Caroll, 1994) into a probabilistic framework. Simulation shows that the proposed method performs better than tandem clustering or clustering without any dimension reduction. When the number of the variables is larger than the number of samples in each cluster, probabilistic reduced K-means show better formation of clusters than non-probabilistic reduced K-means. In the application to a real data set, it revealed similar or better cluster structure compared to other methods.

• Journal of the Korean Data and Information Science Society
• /
• v.21 no.1
• /
• pp.121-128
• /
• 2010

K-means clustering is an iterative algorithm in which items are moved among sets of clusters until the desired set is reached. K-means clustering has been widely used in many applications, such as market research, pattern analysis or recognition, image processing, etc. It can identify dense and sparse regions among data attributes or object attributes. But k-means algorithm requires many hours to get k clusters that we want, because it is more primitive, explorative. In this paper we propose a new method of k-means clustering using a center of gravity for grid-based sample. It is more fast than any traditional clustering method and maintains its accuracy.

• Journal of the Korean Data and Information Science Society
• /
• v.28 no.5
• /
• pp.1133-1144
• /
• 2017

Cluster analysis with meteorological data allows to segment meteorological region based on meteorological characteristics. By the way, meteorological observed data are not adequate for cluster analysis because meteorological stations which observe the data are located not uniformly. Therefore the clustering of meteorological observed data cannot reflect the climate characteristic of South Korea properly. The clustering of $5km{\times}5km$ gridded data derived from a numerical model, on the other hand, reflect it evenly. In this study, we analyzed long-term grid data for temperatures and precipitation using cluster analysis. Due to the monthly difference of climate characteristics, clustering was performed by month. As the result of K-Means cluster analysis is so sensitive to initial values, we used initial values with Ward method which is hierarchical cluster analysis method. Based on clustering of gridded data, cluster of meteorological stations were determined. As a result, clustering of meteorological stations in South Korea has been made spatio-temporal segmentation.

• The Korean Journal of Applied Statistics
• /
• v.34 no.6
• /
• pp.969-977
• /
• 2021

We cluster the electricity consumption of households in A-apartment in Seoul, Korea using Hierarchical K-means clustering algorithm. The data is recorded from the advanced metering infrastructure (AMI), and we focus on the electricity consumption during evening weekdays in summer. Compare to the conventional clustering algorithms, Hierarchical K-means clustering algorithm is recently applied to the electricity usage data, and it can identify usage patterns while reducing dimension. We apply Hierarchical K-means algorithm to the AMI data, and compare the results based on the various clustering validity indexes. The results show that the electricity usage patterns are well-identified, and it is expected to be utilized as a major basis for future applications in various fields.

• The Korean Journal of Applied Statistics
• /
• v.14 no.2
• /
• pp.439-448
• /
• 2001

본 연구는 시간자료(Longitudinal data)의 분석을 위하여 Fuzzy k-means 군집분석 방법을 확장한 알고리즘을 제안한다. 이 논문에서 제안하는 군집분석방법은 각각의 개체에 대응하는 성장곡선에 Fuzzy k-means 군집분석의 알고리즘을 결합하는 것을 핵심아이디어로한다. 분석결과는 생성된 군집을 성장곡선모형으로 표현할 수 있고 또한 추정된 모형의 식을 활용하여 새로운 개체를 분류도 할수 있음을 보인다. 그리고 이 군집분석방법은 아직 자라지 않은 나이 어린 개체가 미래에 어느 군집에 속할 것인가 하는 분류와 함께 이 개체의 향후 성장상태를 예측을 하는 데에도 적용이 가능하다. 제안된 알고리즘을 원숭이(macaque)의 상악동(maxillary sinus)의 자료에 적용한 실례로 보인다.

• Journal of the Korean Data and Information Science Society
• /
• v.28 no.2
• /
• pp.395-406
• /
• 2017

The purpose of this study is to identify the pattern of daily electricity demand through clustering and classification. The hourly data was collected by KPS (Korea Power Exchange) between 2008 and 2012. The time trend was eliminated for conducting the pattern of daily electricity demand because electricity demand data is times series data. We have considered k-means clustering, Gaussian mixture model clustering, and functional clustering in order to find the optimal clustering method. The classification analysis was conducted to understand the relationship between external factors, day of the week, holiday, and weather. Data was divided into training data and test data. Training data consisted of external factors and clustered number between 2008 and 2011. Test data was daily data of external factors in 2012. Decision tree, random forest, Support vector machine, and Naive Bayes were used. As a result, Gaussian model based clustering and random forest showed the best prediction performance when the number of cluster was 8.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.18 no.3
• /
• pp.625-630
• /
• 2014

The amount of information is increasing rapidly with the development of the internet and the computer. Since these enormous information is managed by the document forms, it is necessary to search and process them efficiently. The document clustering technique which clusters the related documents through the similarity between the documents help to classify, search, and process the large amount of documents automatically. This paper proposes a method to find the initial seed points through principal component analysis when the documents represented by vectors in the feature vector space are clustered by K-means algorithm in order to increase clustering performance. The experiment shows that our method has a better performance than the traditional K-means algorithm.

• Journal of the Korean Data and Information Science Society
• /
• v.22 no.4
• /
• pp.741-753
• /
• 2011

We accomplish clustering analyses for yeast cell cycle microarray expression data. We compare model-based clustering, K-means, PAM, SOM and hierarchical Ward method with yeast data. As the validity measure for clustering results, connectivity, Dunn Index and silhouette values are computed and compared.

• The Korean Journal of Applied Statistics
• /
• v.22 no.4
• /
• pp.745-758
• /
• 2009

Cluster analysis is the automated search for groups of related observations in a data set. To group the observations into clusters many techniques has been proposed, and a variety measures aimed at validating the results of a cluster analysis have been suggested. In this paper, we compare complete linkage, Ward's method, K-means and model-based clustering and compute validity measures such as connectivity, Dunn Index and silhouette with simulated data from multivariate distributions. We also select a clustering algorithm and determine the number of clusters of Korean consumers based on Korean consumers' palatability scores for Hanwoo bull in BBQ cooking method.