정부-민간 협력관계를 중심으로 한 클러스터 진화유형분석 : 한국의료산업 사례분석

박기원,최현도 한국혁신학회 2019 한국혁신학회지 Vol.14 No.4

The government and the private sector attempt to create clusters of high-tech industries across regions. Government and business need a strategic approach to encourage cluster development. This study combines cluster theory and government policy formation theory to form three different approaches to cluster evolution and derive relevant propositions by these evolution types. In addition, we will analyze the government and business’s innovation patterns for each cluster and suggest a way to improve the cluster policy. This research goal was achieved by using the industry and patent data provided by various cluster-related organizations in the region, using the Korean medical industry cluster as a case. This study shows the types of evolutionary patterns of clusters and its various performance, which will help the government and local governments to establish cluster development policies, and to set specific directions on how to improve innovation and business models of firms in clusters. 정부와 민간은 각 지역에 첨단산업 기반의 클러스터 형성에 노력을 기울이고 있다. 클러스터 활성화를 위해서 정부와 기업은 전략적인 접근이 필요하다. 본 연구는 클러스터 이론과 정부정책형성 이론을 결합하여 클러스터 육성에 대한 세 가지 다른 접근방법을 유형하고 유형별로 관련 명제를 도출할 것이다. 또한 각 클러스터 유형별 정부와 민간기업 혁신패턴을 분석하여, 향후 정부의 클러스터 정책에 대한 개선방안을 제시할 것이다. 이와 같은 연구목표는 한국의 의료산업클러스터를 사례로 지역의 다양한 클러스터 유관단체에서 제공하는 현황자료와 특허자료를 이용해 달성할 수 있었다. 본 연구는 클러스터 유형별 진화 패턴과 그에 따른 다양한 성과를 보여줌으로써, 정부 및 지방자치단체의 클러스터 육성정책 수립과 클러스터 내 기업의 기술혁신과 비즈니스 모델 혁신 방법에 대한 구체적 방향설정에 도움을 줄 것이다.

Nearest neighbor and validity-based clustering

Seo H. Son,Suk T. Seo,Soon H. Kwon 한국지능시스템학회 2004 INTERNATIONAL JOURNAL of FUZZY LOGIC and INTELLIGE Vol.4 No.3

The clustering problem can be formulated as the problem to find the number of clusters and a partition matrix from a given data set using the iterative or non-iterative algorithms. The author proposes a nearest neighbor and validity-based clustering algorithm where each data point in the data set is linked with the nearest neighbor data point to form initial clusters and then a cluster in the initial clusters is linked with the nearest neighbor cluster to form a new cluster. The linking between clusters is continued until no more linking is possible. An optimal set of clusters is identified by using the conventional cluster validity index. Experimental results on well-known data sets are provided to show the effectiveness of the proposed clustering algorithm.

Clustering the Clinical Course of Chronic Urticaria Using a Longitudinal Database: Effects on Urticaria Remission

예영민,윤지원,Woo Seong-Dae,Jang Jae-Hyuk,Lee Youngsoo,이현영,신유섭,남동호,박해심 대한천식알레르기학회 2021 Allergy, Asthma & Immunology Research Vol.13 No.3

Purpose Little is known about the clinical course of chronic urticaria (CU) and predictors of its prognosis. We evaluated CU patient clusters based on medication scores during the initial 3 months of treatment in an attempt to investigate time to remission and relapse rates for CU and to identify predictors for CU remission. Methods In total, 4,552 patients (57.9% female; mean age of 38.6 years) with CU were included in this retrospective cohort study. The K-medoids algorithm was used for clustering CU patients. Kaplan-Meier survival analysis with Cox regression was applied to identify predictors of CU remission. Results Four distinct clusters were identified: patients with consistently low disease activity (cluster 1, n = 1,786), with medium-to-low disease activity (cluster 2, n = 1,031), with consistently medium disease activity (cluster 3, n = 1,332), or with consistently high disease activity (cluster 4, n = 403). Mean age, treatment duration, peripheral neutrophil counts, total immunoglobulin E, and complements levels were significantly higher for cluster 4 than the other 3 clusters. Median times to remission were also different among the 4 clusters (2.1 vs. 3.3 vs. 6.4 vs. 9.4 years, respectively, P < 0.001). Sensitization to house dust mites (HDMs; at least class 3) and female sex were identified as significant predictors of CU remission. Around 20% of patients who achieved CU remission experienced relapse. Conclusions In this study, we identified 4 CU patient clusters by analyzing medication scores during the first 3 months of treatment and found that sensitization to HDMs and female sex can affect CU prognosis. The use of immunomodulators was implicated in the risk for CU relapse. Purpose Little is known about the clinical course of chronic urticaria (CU) and predictors of its prognosis. We evaluated CU patient clusters based on medication scores during the initial 3 months of treatment in an attempt to investigate time to remission and relapse rates for CU and to identify predictors for CU remission. Methods In total, 4,552 patients (57.9% female; mean age of 38.6 years) with CU were included in this retrospective cohort study. The K-medoids algorithm was used for clustering CU patients. Kaplan-Meier survival analysis with Cox regression was applied to identify predictors of CU remission. Results Four distinct clusters were identified: patients with consistently low disease activity (cluster 1, n = 1,786), with medium-to-low disease activity (cluster 2, n = 1,031), with consistently medium disease activity (cluster 3, n = 1,332), or with consistently high disease activity (cluster 4, n = 403). Mean age, treatment duration, peripheral neutrophil counts, total immunoglobulin E, and complements levels were significantly higher for cluster 4 than the other 3 clusters. Median times to remission were also different among the 4 clusters (2.1 vs. 3.3 vs. 6.4 vs. 9.4 years, respectively, P < 0.001). Sensitization to house dust mites (HDMs; at least class 3) and female sex were identified as significant predictors of CU remission. Around 20% of patients who achieved CU remission experienced relapse. Conclusions In this study, we identified 4 CU patient clusters by analyzing medication scores during the first 3 months of treatment and found that sensitization to HDMs and female sex can affect CU prognosis. The use of immunomodulators was implicated in the risk for CU relapse.

학습분석학 관점의 대학 이러닝 학습자 군집화와 학업성취도 관계 분석 : 이러닝 학습 시. 공간 데이터를 기반으로

이해듬 한국방송통신대학교 미래원격교육연구원 2018 평생학습사회 Vol.14 No.3

This study was designed to approach various e-Learning data accumulated in Learning Management System (LMS) for university e-Learning from the perspective of learning analytics. This study used cluster analysis method with observation variables being e-Learning spatio-temporal data and analyzed the differences in academic achievement among clusters. For this study researcher collected e-Learning data from 68 e-Learning classes, 13,611 learners, during 3 years (6 semesters). This study used cluster analysis from spatio-temporal data, found out differences between attendance rate and used F-test to find out different academy achievement. Major study findings were as follow: Firstly, the number of clusters of university e-Learning learners emerged four (Cluster 1-4). Characteristics of each cluster were classified as [Cluster 1] of learners mainly outside school-weekdays-in the afternoon, [Cluster 2] those in school-weekdays-in the afternoon, [Cluster 3] outside school-weekends-in the afternoon and [Cluster 4] outside school-weekdays-at night. Secondly, Cluster 1 and Cluster 2 showed higher attendance than Cluster 3, Cluster 4 with both having statistical significance (F=68.34, p<.001). Also Cluster 2 and Cluster 1 received higher academic achievement than Cluster 3, Cluster 4 with both having statistical significance (F=39.60, p<.001). 본 연구는 대학 이러닝 학습관리시스템(LMS)에 축적된 다양한 학습데 이터를 학습분석학 관점에서 접근하기 위해 이러닝 강좌를 수강하고 있는 학습자의 학습 시・공간 데이터를 기반으로 학습패턴을 군집화하여 학업 성취도 간에 의미 있는 차이를 규명하였다. 이를 위해 3년간(6학기) 68개 이러닝 강좌의 수강생 1만 3,611명의 이러닝 학습데이터를 표집하였고, 자 료분석은 학습 시.공간 데이터에 의한 군집분석과 군집 간 출석률, 학업 성취도 차이 분석의 검증을 수행하였다. 본 연구의 주요 연구 결과를 요 약하면 다음과 같다. 첫째, 이러닝 학습자의 학습공간(교내, 교외)과 학습 시간대(오전-오후-야간, 평일-주말)의 ‘학습환경 데이터’에 기초한 7개 변 인을 투입하여 학습패턴의 군집분석을 수행한 결과 군집의 수는 4개(군집 1~군집 4)로 추출되었다. 군집별 특성을 요약하면 [군집 1] 교외-평일-오 후시간대 주학습자, [군집 2] 교내-평일-오후시간대 주학습자, [군집 3] 교 외-주말-오후시간대 주학습자, [군집 4] 교외-평일-야간시간대 주학습자로 나타났다. 이러닝 학습자의 학습패턴 군집에 따른 학업성과 차이를 분석한 결과 출석률에서는 [군집 1] 학습자와 [군집 2] 학습자(=92.01)가 더 높 게 나타났으며, 차이검증에서도 통계적으로 유의미( =68.34, <.01)하게 분석되었다. 학업성취도에서는 [군집 2] 학습자(=74.29)와 [군집 1] 학습 자가 다른 군집보다 더 높은 학업성취도를 보여 주었고, 통계적으로도 유 의미한 차이가 있는 것( =39.60, <.001)으로 분석되었다. 전반적으로 평 일-오후시간대에 주로 학습하는 [군집 2]와 [군집 1]의 이러닝 학습자는 매주 규칙적인 학습패턴으로 이러닝을 학습하고, 이는 출석률과 학업성취 도에 유의한 영향을 주는 것으로 추론할 수 있다.

Priorities, Mechanisms and Prospects on Industrial Clusters and Special Economic Zones in Kazakhstan

Yespayev, Saken S. Korea Distribution Science Association 2014 The Journal of Asian Finance, Economics and Busine Vol.1 No.2

This research investigates the characteristics, principles, advantages, factors and problems of cluster development in Kazakhstan, and identifies the prerequisites, conditions and stages of organizing clusters on the framework of special economic zones. In this research, we used methods, which will allow analyzing of the organization industrial clusters in special economic zones in Kazakhstan. The author studied international experience of cluster development and the efficiency of the use of the model of the "rhombus effect" with account the specific features of interaction between the participants of the cluster, analysis of the legal framework for the formation and development of clusters. These have been identified as the more important or strategically necessary clusters in Kazakhstan: innovation-technological cluster, innovation-education cluster, innovation-petrochemical cluster, innovative-metallurgical cluster, transport and logistics cluster, textile industry cluster, tourism cluster, agro cluster, construction cluster, medical and pharmaceutical cluster. Firstly, the results suggest that the interaction of science, education, business and government in the development and implementation of innovation policy is not sufficiently structured to provide a balanced representation of the interests of the range of various innovative enterprises in Kazakhstan. Secondly, the legal basis of cluster development in Kazakhstan is determined. Need to develop mechanisms for the implementation of promising direction. Thirdly, the clusters can be formed in the existing special economic zones, allowing them to get right to the mass production of high-tech products that are developed.

유아 동기 유형 평가도구 적용 군집화 및 사례분석

유구종 ( Yoo Ku Jong ),최승연 ( Choi Seung Yeon ),조희정 ( Cho Hee Jung ),성은영 ( Sung Eun Young ) 한국어린이문학교육학회 2017 어린이문학교육연구 Vol.18 No.1

본 연구는 유아 동기 유형 평가도구에 의해 평가가 실시되었을 때 유아의 동기 유형이 어떠한 하위 군집의 유형으로 분류 되는지 살펴보고 구체적인 사례를 통해 하위군집의 특성을 살펴봄으로써 현장 적용성을 비롯한 생태학적 타당도를 제고하는데 그 목적이 있다. 이를 위해 본 연구에서는 614명의 유아를 대상으로 유아 동기 유형 평가도구를 적용하여 군집분석을 실시하여 군집화 된 군집유형을 명명하고, 성별과 연령에 따라 군집 유형 차이를 검증하였다. 또한 만 5세 유아 20명으로 구성된 1개 학급을 대상으로 군집화를 통해 도출한 군집 유형으로 연구대상을 유형별로 나누어 군집유형에 따른 행동 특성을 파악하고자 사례분석을 실시하였다. 이를 위해 계층적 군집분석, K-means 군집분석의 총 2회에 걸친 군집분석을 통해 3개의 군집으로 유아의 동기 유형을 군집화 하였다. 군집 1은 동기 유형에 있어 내적 동기가 월등히 높고 외적 동기와 무동기는 매우 낮은 것으로 나타나 고(高)내적 동기형으로 명명되었고, 군집 2는 내적 동기가 군집 1과 군집 3에 비하여 절반수준에 머물렀고, 외적 동기와 무동기의 경우에도 군집 3에 비해 저조한 수준으로 나타나 저(低)동기형으로 명명하였다. 군집 3의 경우는 내적 동기는 군집 1과 비슷한 수준으로 나타났으나 무동기 유형이 가장 높게 나타나 복합동기형으로 명명하였다. 성별과 연령에 따른 차이분석은 모두 차이가 없는 것으로 나타나 동기는 연령과 성별에 관계없는 개인의 일반적인 특성인 것으로 나타났다. 사례분석에서는 군집분석을 통해 군집화 된 군집별 특징이 사례에 부합되어 발현되는 것으로 나타났다. For clustering of infant motivation types, hierarchical cluster analysis and K-means cluster analysis were conducted. As a result, infant motivation types were classified into 3 clusters. As for the motivation types of cluster 1, internal motivation was significantly outstanding while external motivation and internal motivation were quite insignificant. Internal motivation of cluster 2 was as low as a half of those of cluster 1 and cluster 3. External motivation and no-motivation as well were lower than those of cluster 3. Internal motivation of cluster 3 was similar to that of cluster 1, but the level of no-motivation was the highest. Accordingly, in this study, cluster 1 is called the high internal motivation type, cluster 2 the low motivation type, and cluster 3 the complex motivation type, respectively, to represent the attributes of the clusters. As a result of analyzing the differences among the cluster types depending on the sex and age, it was determined that there was no significant difference depending on the sex and age of the infant. As a case study on the clusters of motivation types, anecdote recordings were conducted for 12 weeks and analyzed systematically. As a result, it was determined that young children in the high internal motivation type cluster were seeking self-satisfaction and satisfying their internal curiosity while young children in the low motivation type were following the instruction of others and expected extrinsic rewards. Young children in the complex motivation type cluster were sensitive to external stimuli and would act at a moderate level.

Emergence of East Asian TFT-LCD Clusters: A Comparative Analysis of the Samsung Cluster in South Korea and the Chimei Cluster in Taiwan

윤진효,박상문,임동욱,함성득 기술경영경제학회 2010 ASIAN JOURNAL OF TECHNOLOGY INNOVATION Vol.18 No.1

This paper investigates cluster formation and the development processes of new thin file transistor liquid crystal display (TFT-LCD) clusters in East Asia. Despite the pivotal role of clusters in regional development and national competitiveness, there are only a few studies on how new East Asian high-tech clusters have emerged and evolved and how these clusters are similar to and different from other clusters. Based on a comparative analysis of new TFT-LCD clusters between Samsung in Asan-Tangjung, South Korea, and Chimei in Tainan, Taiwan, we examine dynamic development processes and investigate how these rural areas have changed into high-tech clusters in only a decade’s time. Specifically, this paper explores the preconditions and initiation characteristics of TFT-LCD clusters. It also compares some similarities and differences between two East Asian TFT-LCD clusters and investigates the uniqueness of other global clusters. Therefore, this paper enhances our understanding of the dynamics of industrial clusters, adds a comparative perspective on cluster analysis, and suggests policy implications from the case study of cluster formation in South Korea and Taiwan.

주파수 및 시간 특성을 활용한 머신러닝 기반 공동주택 주거소음의 군집화 및 분류

김정훈,이송미,김수홍,송은성,류종관 한국음향학회 2023 韓國音響學會誌 Vol.42 No.6

본 연구는 주파수 및 시간 특성을 활용하여 머신러닝 기반 공동주택 주거소음의 군집화 및 분류를 진행하였다. 먼저, 공동주택 주거소음의 군집화 및 분류를 진행하기 위하여 주거소음원 데이터셋을 구축하였다. 주거소음원 데이터셋은 바닥충격음, 공기전달음, 급배수 및 설비소음, 환경소음, 공사장 소음으로 구성되었다. 각 음원의 주파수 특성은 1/1과 1/3 옥타브 밴드별 Leq와 Lmax값을 도출하였으며, 시간적 특성은 5 s 동안의 6 ms 간격의 음압레벨 분석을 통해Leq값을 도출하였다. 공동주택 주거소음원의 군집화는 K-Means clustering을 통해 진행하였다. K-Means의 k의 개수는 실루엣 계수와 엘보우 방법을 통해 결정하였다. 주파수 특성을 통한 주거소음원 군집화는 모든 평가지수에서 3개로군집되었다. 주파수 특성 기준으로 분류된 각 군집별 시간적 특성을 통한 주거소음원 군집화는 Leq평가지수의 경우 9 개, Lmax 경우는 11개로 군집되었다. 주파수 특성을 통해 군집된 각 군집은 타 주파수 대역 대비 저주파 대역의 음에너지의 비율 또한 조사되었다. 이후, 군집화 결과를 활용하기 위한 방안으로 세 종류의 머신러닝 방법을 이용해 주거소음을 분류하였다. 주거소음 분류 결과, 1/3 옥타브 밴드의 Leq값으로 라벨링된 데이터에서 가장 높은 정확도와 f1-score 가 나타났다. 또한, 주파수 및 시간적 특성을 모두 사용하여 인공신경망(Artificial Neural Network, ANN) 모델로 주거소음원을 분류했을 때 93 %의 정확도와 92 %의 f1-score로 가장 높게 나타났다. In this study, machine learning-based clustering and classification of residential noise in apartment buildings was conducted using frequency and temporal characteristics. First, a residential noise source dataset was constructed . The residential noise source dataset was consisted of floor impact, airborne, plumbing and equipment noise, environmental, and construction noise. The clustering of residential noise was performed by K-Means clustering method. For frequency characteristics, Leq and Lmax values were derived for 1/1 and 1/3 octave band for each sound source. For temporal characteristics, Leq values were derived at every 6 ms through sound pressure level analysis for 5 s. The number of k in K-Means clustering method was determined through the silhouette coefficient and elbow method. The clustering of residential noise source by frequency characteristic resulted in three clusters for both Leq and Lmax analysis. Temporal characteristic clustered residential noise source into 9 clusters for Leq and 11 clusters for Lmax. Clustering by frequency characteristic clustered according to the proportion of low frequency band. Then, to utilize the clustering results, the residential noise source was classified using three kinds of machine learning. The results of the residential noise classification showed the highest accuracy and f1-score for data labeled with Leq values in 1/3 octave bands, and the highest accuracy and f1-score for classifying residential noise sources with an Artificial Neural Network (ANN) model using both frequency and temporal features, with 93 % accuracy and 92 % f1-score.

How Firms Develop Linkages for Development and Growth: Cases in Malaysian Greenfield and Brownfield Technology Parks

Avvari V. Mohan,Isshamudin Ismail 세계과학도시연합 2015 World Technopolis Review Vol.4 No.2

This paper aims to explore how firms develop and grow in regional clusters based in a developing country. The argument is that start-ups / small and large firms are able to grow by developing linkages or networks for resources within clusters – and this tenet is based on studies of firms that are based from such clusters as Silicon Valley in the US, Cambridge in UK and other clusters from which have evolved over long periods of time. Most of the time we hear narratives from the developed world where there are brownfield cluster development efforts. In developing countries governments are making efforts to develop clusters from scratch – which in this paper we term as greenfield cluster versus a brownfield development, which is where the cluster is developed based on existing and new organisations in a region. In this paper, we believe the context of clusters can be important in determining the way firms develop linkages for their growth – and we look at two contexts in Malaysia ie. A greenfield cluster and a brownfield cluster. The paper presents findings from case studies of firms in a greenfield cluster (Cyberjaya) and a brown field cluster (Penang) in Malaysia. The cases reveal fairly different approaches to development of linkages or networks, which we hope will provides insights to cluster development officials and policy makers and implications to researchers for developing studies of clusters and innovation systems.

A sequential clustering algorithm with applications to gene expression data

송종우,Dan L. Nicolae 한국통계학회 2009 Journal of the Korean Statistical Society Vol.38 No.2

Clustering algorithms are used in the analysis of gene expression data to identify groups of genes with similar expression patterns. These algorithms group genes with respect to a predefined dissimilarity measure without using any prior classification of the data. Most of the clustering algorithms require the number of clusters as input, and all the objects in the dataset are usually assigned to one of the clusters. We propose a clustering algorithm that finds clusters sequentially, and allows for sporadic objects, so there are objects that are not assigned to any cluster. The proposed sequential clustering algorithm has two steps. First it finds candidates for centers of clusters. Multiple candidates are used to make the search for clusters more efficient. Secondly, it conducts a local search around the candidate centers to find the set of objects that defines a cluster. The candidate clusters are compared using a predefined score, the best cluster is removed from data, and the procedure is repeated. We investigate the performance of this algorithm using simulated data and we apply this method to analyze gene expression profiles in a study on the plasticity of the dendritic cells.