• Annual Conference of KIPS
• /
• 2014.04a
• /
• pp.952-955
• /
• 2014

본 논문에서는 정보 검색 분야에서 잘 알려진 잠재 의미 분석 방법과 계층적 군집화 방법의 단점을 상호 보완하여 보다 효율적인 정보 검색을 위한 혼합형 군집화 방법을 제안한다. 먼저, 잠재 의미 분석 방법은 벡터 연산을 통하여 자동적으로 문서 내에 있는 잠재적인 의미를 찾는 정보 검색분야에서 많이 사용되는 고전적인 방법이다. 그러나 이 방법은 언어의 유의성이나 다의성으로 인하여 발생되는 백-오브-워드(bag-of-word) 문제를 가지고 있다. 두 번째 방법인 문서 군집화를 위하여 범용적으로 사용되고 있는 계층적 군집화 방법이다. 이 방법은 이를 통하여 분석된 군집의 질적 측면에서 볼 때, 여전히 단층적 군집들이 많이 형성되어 세부적인 분석을 통한 추가적인 군집화가 필요함을 알 수 있다. 따라서, 본 논문에서는 앞서 언급한 문제점을 해결하기 위하여 혼합적인 방법으로 잠재 의미 분석 방법을 이용한 응집 계층 군집화 방법을 제안한다. 제안한 방법을 이용하여 잘 알려진 두 개의 데이터에 적용하고 기존의 방법과 그 결과를 비교함으로써 군집의 질적 측면에서의 우수함을 보인다.

• Journal of Digital Convergence
• /
• v.19 no.12
• /
• pp.557-565
• /
• 2021

This study was attempted to identify differences in self-esteem and grit in adolescents depending on the type of parenting attitude. Among the Korea Children Youth Panel Survey conducted by National Youth Policy Institute, the data of 2,438 first-year middle school students in 2018 year were analyzed. The collected data were analyzed using hierarchical cluster analysis and k-mean cluster analysis. As a result, the adolescent's perceived parenting attitude was classified into four types: 'passive affection acceptance', 'active affection acceptance', 'authoritarian inconsistency', and 'lack of affection rejection'. Also, there were significant differences in self-esteem and the degree of grit among the four clusters of parenting attitudes. Both self-esteem and grit were highest in the "active affection acceptance" group 2. In the future, differentiated parental education is needed for each cluster to improve self-esteem and grit of adolescents, and this study can be used as a basic data for the development of educational programs.

• The Korean Journal of Applied Statistics
• /
• v.35 no.3
• /
• pp.407-419
• /
• 2022

The semiconductor fabrication process is complex and time-consuming. There are sometimes errors in the process, which results in defective die on the wafer bin map (WBM). We can detect the faulty WBM by finding some patterns caused by dies. When one manually seeks the failure on WBM, it takes a long time due to the enormous number of WBMs. We suggest a two-step approach to discover the probable pattern on the WBMs in this paper. The first step is to separate the normal WBMs from the defective WBMs. We adapt a hierarchical clustering for de-noising, which nicely performs this work by wisely tuning the number of minimum points and the cutting height. Once declared as a faulty WBM, then it moves to the next step. In the second step, we classify the patterns among the defective WBMs. For this purpose, we extract features from the WBM. Then machine learning algorithm classifies the pattern. We use a real WBM data set (WM-811K) released by Taiwan semiconductor manufacturing company.

• The Journal of Korean Association of Computer Education
• /
• v.22 no.5
• /
• pp.51-65
• /
• 2019

This study aims to provide educational implications for more strategic online software education by the types of online learning according to learners' self-regulated learning characteristics in the online software education environment and examining the characteristics of each type. For this, variables related to self-regulated learning characteristic were extracted from the log data of 809 students participating in the online software learning program of K University, and then analyzed using hierarchical cluster analysis. Based on hierarchical cluster analysis learner clusters according to the characteristics of self-regulated learning were derived and the differences between learners' learning characteristics and learning results according to cluster types were examined. As a result, the types of self-regulated learning of online software learners were classified as 'high level self-regulated learning type (group 1)', 'medium level self-regulated learning type (group 2)', and 'low level self-regulated learning type (group 3)'. The achievement level was found to be highest in 'high-level self-regulated learning type (group 1)' and 'low-level self-regulated learning type (group 3)' was the lowest. Based on these results, the implications for effective online software education were suggested.

• Journal of Korean Society of Transportation
• /
• v.32 no.6
• /
• pp.615-627
• /
• 2014

In this study, nonhierarchical K-mean cluster analysis is used to classify the types of 230 local governments and the Mann-Whitney U test and Kruskal-Wallis analysis are used to analyze the characteristics of pedestrian accidents by region types. With empirical analysis of pedestrian accidents, this study suggests improvements of walking environments reflecting local characteristics. Type 1-A (relatively dominant urban commercial areas), Type 1-B (predominantly urban residence) and Type 2 (rural areas) have been classified using nonhierarchical K-mean cluster analysis. According to the results, pedestrian accident rate on community roads was more than 60% for all types and incidence rate in rural areas was higher than that in urban areas. In addition, pedestrian accidents of Type 1-B have been found to occur more frequently than Type 2 in intersections and crossings, while the number of roadside casualties for Type 2 was highest.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2008.05a
• /
• pp.1072-1076
• /
• 2008

본 연구에서는 한강유역 109개 지점의 강우관측소에서 관측된 지속기간별 연최대강우량을 기본으로 각 지속기간별 L-모멘트값을 산정하고, 한강유역에 적합한 빈도해석기법을 정의하기 위하여 지역구분을 실시하였다. 지역구분을 위한 군집분석을 수행하기 위하여 각 지점별 기상학적 인자와 지형학적 인자를 변수로 사용하였다. 군집분석 기법인 Ward, 평균연결법, Fuzzy-c means, Two-Step방법을 이용하여 지역구분을 실시하였다. GIS를 이용하여 각 방법들을 이용하여 군집된 결과를 도시한 결과 Fuzzy-c means방법으로 구분된 지역구분이 적합한 것으로 나타났다. 또한 구분된 지역의 동질성 여부를 판단하고 적정 분포형을 선정하였으며 지점빈도해석 및 지역빈도해석을 통하여 빈도별 확률 수문량을 산정하였다. 산정된 결과의 정확도 알아보기 위해 모의발생을 시킨 후, 각 기법별로 산정된 상대 평균 제곱근 오차(Relative Root Mean Square Error, RRMSE)를 비교 분석한 결과 대체적으로 지수홍수법과 계층적 방법이 낮은 RRMSE를 나타냈다. 따라서 한강유역에서는 지수홍수법과 계층적 방법을 적용한 지역빈도해석이 적합한 것으로 판단된다.

• The KIPS Transactions:PartB
• /
• v.10B no.1
• /
• pp.33-40
• /
• 2003

Data mining plays an important role in a knowledge discovery process and various algorithms of data mining can be selected for the specific purpose. Most of traditional hierachical clustering methode are suitable for processing small data sets, so they difficulties in handling large data sets because of limited resources and insufficient efficiency. In this study we propose a hybrid neural networks clustering technique, called PPC for Pre-Post Clustering that can be applied to large data sets and find unknown patterns. PPC combinds an artificial intelligence method, SOM and a statistical method, hierarchical clustering technique, and clusters data through two processes. In pre-clustering process, PPC digests large data sets using SOM. Then in post-clustering, PPC measures Similarity values according to cohesive distances which show inner features, and adjacent distances which show external distances between clusters. At last PPC clusters large data sets using the simularity values. Experiment with UCI repository data showed that PPC had better cohensive values than the other clustering techniques.

• The Journal of Korean Association of Computer Education
• /
• v.23 no.1
• /
• pp.77-90
• /
• 2020

The purpose of the study is to provide educational implications for more effective Problem-based learning(PBL) by investigating students' learning types based on their online learning behaviors. A total of 1,341 students participated in the study, and they engaged in a six-week-long PBL program run by K University. For the study, participants' online activity data were collected. From the data, a total of 48 variables that represent their various online learning behaviors were extracted. Based on the variables, hierarchical cluster analysis was conducted to analyze learning types. Also, the differences in learning characteristics and achievements were investigated by considering types of learning. As a result, the learning types in online PBL were classified as 'high-level participation (cluster 1)', 'medium-level participation (cluster 2)', and 'low-level participation (cluster 3)'. In addition, the achievement level was found to be highest in 'high-level participation (cluster 1)' and lowest in 'low-level participation (cluster 3)'. Based on the results, the implications for improving online PBL were suggested.

• Culinary science and hospitality research
• /
• v.18 no.4
• /
• pp.47-58
• /
• 2012

This study, targeting the students of "K" university in Busan City area, was performed to draw the groups by food-related lifestyle types and to identify the correlation between each group's attributes of selecting places to eat out and obesity index. The purpose of the study was achieved by means of the PASW Statistic 18.0(Predictive Analytics Software) which conducted frequency analysis, factor analysis, reliability analysis, t-test, ${\chi}^2$-test, non-hierarchical cluster analysis and ANOVA. It turned out that the male university students were 175.59 cm tall and weigh 69.53 kg on average. And the female university students showed their average height of 162.81 cm and weight of 53.42 kg. When examined by the body mass index(BMI), male students were composed of 1.7% of underweight, 64.6% of normal weight, 19.7% of overweight and 14.0% of obese. As for the female students, 22.9% were classified as underweight, 62.7% as normal weight, 8.5% as overweight and 5.9% as obese. The food-related lifestyle categories were divided into five factors; health seeking type, safety seeking type, mood seeking type, taste seeking type, and western food seeking type. The four attributes of selecting places to eat out included quality of food and service, price reasonableness, accessibility and atmosphere, and experience to have eaten. With regard to food-related lifestyle, the groups were named by cluster 1 [careless diet group], Cluster 2 [health oriented group], and cluster3 [careless healthcare group]. In terms of the correlation between the clusters by food-related lifestyle and their attributes of selecting places to eat out, Cluster 1 had a high mean value in experience to have eaten, Cluster 2 quality of food and service, Cluster 3 accessibility and atmosphere.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.05a
• /
• pp.447-449
• /
• 2021

Various wireless sensor network protocols have been proposed to maintain the network for a long time by minimizing energy consumption. Using the K-means clustering algorithm takes longer to cluster than traditional hierarchical algorithms because the center point must be moved repeatedly until the final cluster is established. For K-means clustering-based protocols, only the residual energy of nodes or nodes near the center point of the cluster is considered when the cluster head is elected. In this paper, we propose a new wireless sensor network protocol based on K-means clustering to improve the energy efficiency while improving the aforementioned problems.