하이브리드 검색알고리즘을 이용한 K-평균 군집화의 성능향상

전성해 청주대학교 2007 産業科學硏究 Vol.25 No.1

In this paper, we propose a hybrid searching algorithm for K-means clustering. This is a clustering method based on sample standard error. Generally the number of clusters is important part in K-mean clustering algorithm. Many researchers have determined the number subjectively by the art of prior experience. But, we can not guarantee the number of clusters based on subjective selection. So, we use a hybrid searching method to settle the problem of traditional K-means clustering. To verify improved performance of our work, we make experiments using three data sets by simulation examples.

통계적 학습이론을 이용한 효율적인 데이터 마이닝 전략

전성해 청주대학교 산업과학연구소 2007 産業科學硏究 Vol.24 No.2

Statistical Learning theory(SLT) by Vapink was introduced to overcome the problems of local optima and over-fitting in machine learing algorithms. SLT is consisted of support vector machine(SVM), support vector regression(SVR), and support vector clustering(SVC) for classification, regression, and clustering respectively. So, SLT is a powerful tool for supervised and unsupervised learning at once. There is no algorithm to support supervised and unsupervised tools at a time. But, SLT is able to contribute two learning algorithms. In this paper, we propose an efficient data mining approach using SLT. To verify our work, we use data set from UCI machine learning repository and R-project.

통계적 학습 이론을 이용한 전자 상거래의 추천 시스템

전성해 청주대학교 산업과학연구소 2004 産業科學硏究 Vol.21 No.2

According as the internet increases, the marketing processes are depended on the web. All kinds of goods and sales which are traded on the internet shopping malls are extremely increases. So, the necessity of automatic information system is shown, this system manages to web site connected users. For the recommendation system which can offer user a fit information from numerous web contents. In this paper, the automatic recommendation system which furnish necessary information to connected web user using statistical learning theory is shown. The performance of proposed recommendation system is compared with the existing methods by real data of the actual shopping web site. The predictive accuracy of the proposed system IS improved by comparison with others.

개념 서술 테이터 마이닝을 이용한 고객 세분화 모형

전성해 청주대학교 2004 産業科學硏究 Vol.22 No.1

Data mining technology has been used in many areas and it has been supported by Statistics, database management system, dataware, and machine learning. The works of data mining may be very difficult. Because the data sets of mining process are very large and complex for analysis. So many studies have been researched in the academic and the industry. They have applied data mining to customer relationship management, bioinformatics, artificial intelligence, and so forth. In this paper, we applied it to customer segmentation of Customer Relationship Management using a concept descriptive data mining. This was simpler and better than previous data mining models in customer segmentation. We verified our model with comparative models using the data set from UCI machine learning repository.

기술예측을 위한 통계적 분석기법에 관한 연구

전성해 청주대학교 2013 産業科學硏究 Vol.30 No.2

Most companies have built their R&D planning using technology forecasting results. So, technology forecasting is important to a company. Technology forecasting is to find the future states of target technology. Traditional technology forecasting was based on the experts' experience and knowledge subjectively. So, this forecasting result may be unstable and fluctuated. Statistical analysis is an objective approach based on data for forecasting. In this paper, we study on statistical approach for objective forecasting of target technology.

요인분석을 이용한 최적 군집화

전성해 청주대학교 산업과학연구소 2006 産業科學硏究 Vol.23 No.2

Clustering is an unsupervised learning approach for machine learning. According to diverse reasons, the outliers occur in the clustering. So, the cluster analysis has been a difficult problem. Many researches have been studied on the clustering with outliers. But the good clustering methods have not been yet. In this paper, we proposed an optimal clustering using factor analysis to settle this problem. We verify the performance fo our method using the experimental results by UCI machine learning repository.

Improvement of SOM using Stratification

전성해 한국지능시스템학회 2009 INTERNATIONAL JOURNAL of FUZZY LOGIC and INTELLIGE Vol.9 No.1

Self organizing map(SOM) is one of the unsupervised methods based on the competitive learning. Many clustering works have been performed using SOM. It has offered the data visualization according to its result. The visualized result has been used for decision process of descriptive data mining as exploratory data analysis. In this paper we propose improvement of SOM using stratified sampling of statistics. The stratification leads to improve the performance of SOM. To verify improvement of our study, we make comparative experiments using the data sets form UCI machine learning repository and simulation data.

데이터정보기술의 효율적인 관리방안

전성해 청주대학교 2011 産業科學硏究 Vol.29 No.1

Information technology has been developed rapidly and widely. This has affected all fields of human life. Modern technologies of software and hardware have made remarkable progress. But, dataware has been not developed actively. Whether or not information system succeeds depends on the dataware. In this paper, we study on the effective management of data information technology for traditional information system. Patent documents related to data information technology will be used for verifying our proposed work.

Improvement of Self Organizing Maps using Gap Statistic and Probability Distribution

전성해 한국지능시스템학회 2008 INTERNATIONAL JOURNAL of FUZZY LOGIC and INTELLIGE Vol.8 No.2

Clustering is a method for unsupervised learning. General clustering tools have been depended on statistical methods and machine learning algorithms. One of the popular clustering algorithms based on machine learning is the self organizing map(SOM). SOM is a neural networks model for clustering. SOM and extended SOM have been used in diverse classification and clustering fields such as data mining. But, SOM has had a problem determining optimal number of clusters. In this paper, we propose an improvement of SOM using gap statistic and probability distribution. The gap statistic was introduced to estimate the number of clusters in a dataset. We use gap statistic for settling the problem of SOM. Also, in our research, weights of feature nodes are updated by probability distribution. After complete updating according to prior and posterior distributions, the weights of SOM have probability distributions for optimal clustering. To verify improved performance of our work, we make experiments compared with other learning algorithms using simulation data sets.

Support Vector Machine based on Stratified Sampling

전성해 한국지능시스템학회 2009 INTERNATIONAL JOURNAL of FUZZY LOGIC and INTELLIGE Vol.9 No.2

Support vector machine is a classification algorithm based on statistical learning theory. It has shown many results with good performances in the data mining fields. But there are some problems in the algorithm. One of the problems is its heavy computing cost. So we have been difficult to use the support vector machine in the dynamic and online systems. To overcome this problem we propose to use stratified sampling of statistical sampling theory. The usage of stratified sampling supports to reduce the size of training data. In our paper, though the size of data is small, the performance accuracy is maintained. We verify our improved performance by experimental results using data sets from UCI machine learning repository.