• The Korean Journal of Applied Statistics
• /
• 제26권2호
• /
• pp.361-373
• /
• 2013

There are a wide variety of clustering algorithms; subsequently, we need a measure of similarity between two clustering methods. Such a measure can compare how well different clustering algorithms perform on a set of data. More numbers of compared clustering algorithms allow for more number of valuers for a measure of similarity between two clustering methods. Thus, we need a simple tool that presents the many values of a measure of similarity to compare many clustering methods. We suggest some graphical tools to compareg many clustering methods.

• Journal of the Korean Institute of Intelligent Systems
• /
• 제15권5호
• /
• pp.521-526
• /
• 2005

The fuzzy entropy is proposed for measuring of uncertainty with the help of relation between distance measure and similarity measure. The proposed fuzzy entropy is constructed through a distance measure. In this study, Hamming distance measure is employed for a distance measure. Also a similarity measure is constructed through a distance measure for the measure of similarity between fuzzy sets or crisp sets and the proposed fuzzy entropies and similarity measures are proved.

• Journal of the Korean Data and Information Science Society
• /
• 제28권1호
• /
• pp.49-56
• /
• 2017

Among the techniques for analyzing big data, the association rule mining is a technique for searching for relationship between some items using various relevance evaluation criteria. This associative rule scheme is based on the direction of rule creation, and there are positive, negative, and inverse association rules. The purpose of this paper is to investigate the applicability of various types of relatively causal strength measures to the types of association rules from the point of view of interestingness measure. We also clarify the relationship between various types of confidence measures. As a result, if the rate of occurrence of the posterior item is more than 0.5, the first measure ($RCS_{IJ1}$) proposed by Good (1961) is more preferable to the first measure ($RCS_{LR1}$) proposed by Lewis (1986) because the variation of the value is larger than that of $RCS_{LR1}$, and if the ratio is less than 0.5, $RCS_{LR1}$ is more preferable to $RCS_{IJ1}$.

• KIPS Transactions on Software and Data Engineering
• /
• 제6권1호
• /
• pp.23-28
• /
• 2017

Shape from focus (SFF) is a technique used to reconstruct 3D shape of objects from a sequence of images obtained at different focus settings of the lens. In this paper, a new shape from focus method for 3D reconstruction of microscopic objects is described, which is based on gradient operator in Mathematical Morphology. Conventionally, in SFF methods, a single focus measure is used for measuring the focus quality. Due to the complex shape and texture of microscopic objects, single measure based operators are not sufficient, so we propose morphological operators with multi-structuring elements for computing the focus values. Finally, an optimal focus measure is obtained by combining the response of all focus measures. The experimental results showed that the proposed algorithm has provided more accurate depth maps than the existing methods in terms of three-dimensional shape recovery.

• The Journal of the Acoustical Society of Korea
• /
• 제27권3호
• /
• pp.119-128
• /
• 2008

This paper analyzes the relationship between two representative statistical methods for bandwidth extension (BWE): Gaussian Mixture Model (GMM) and Hidden Markov Model (HMM) ones, and compares their performances. The HMM method is a memory-based system which was developed to take advantage of the inter-frame dependency of speech signals. Therefore, it could be expected to estimate better the transitional information of the original spectra from frame to frame. To verify it, a dynamic measure that is an approximation of the 1st-order derivative of spectral function over time was introduced in addition to a static measure. The comparison result shows that the two methods are similar in the static measure, while, in the dynamic measure, the HMM method outperforms explicitly the GMM one. Moreover, this difference increases in proportion to the number of states of HMM model. This indicates that the HMM method would be more appropriate at least for the 'blind BWE' problem. On the other hand, nevertheless, the GMM method could be treated as a preferable alternative of the HMM one in some applications where the static performance and algorithm complexity are critical.

• Proceedings of the Korean Statistical Society Conference
• /
• 한국통계학회 2004년도 학술발표논문집
• /
• pp.49-54
• /
• 2004

본 연구에서는 표본 농가의 교체나 무응답으로 인한 데이터의 손실을 최소화하기 위하여 핫덱방법을 적용할 때 필요한 무응답 대체군 형성 방안을 제안하였다. 농가경제조사의 무응답 현황과 특성을 살펴보고, 대체군 형성 방안들을 비교할 수 있는 측도를 제안하였다. 제안된 비교 측도를 이용해서 대체군 형성 방안들을 비교하였다.

• Journal of the Korean Data and Information Science Society
• /
• 제25권4호
• /
• pp.857-868
• /
• 2014

Data mining is the representative analysis methodology in the era of big data, and is the process to analyze a massive volume database and summarize it into meaningful information. Association rule technique finds the relationship among several items in huge database using the interestingness measures such as support, confidence, lift, etc. But these interestingness measures cannot be used to establish a causality relationship between antecedent and consequent item sets. Moreover, we can not know association direction by them. This paper propose causally confirmed association thresholds to compensate for these problems, and then check the three conditions of interestingness measures. The comparative studies with basic association thresholds, causal association thresholds, and causally confirmed association thresholds are shown by simulation studies. The results show that causally confirmed association thresholds are better than basic and causal association thresholds.

• Journal of the Korean Data and Information Science Society
• /
• 제28권3호
• /
• pp.515-524
• /
• 2017

In this study, a measure of discrepancy based on MV (margin of victory) has been suggested that might be useful in determining the size of random forest for classification. Here MV is a scaled difference in the votes, at infinite random forest, of two most popular classes of current random forest. More specifically, max(-MV,0) is proposed as a reasonable measure of discrepancy by noting that negative MV values mean a discrepancy in two most popular classes between the current and infinite random forests. We propose an appropriate diagnostic statistic based on this measure that might be useful for the determination of random forest size, and then we derive its asymptotic distribution. Finally, a simulation study has been conducted to compare the performances, in finite samples, between this proposed statistic and other recently proposed diagnostic statistics.

• Journal of the Korean Data and Information Science Society
• /
• 제25권2호
• /
• pp.365-371
• /
• 2014

Association rule of the well-studied techniques in data mining is the exploratory data analysis for understanding the relevance among the items in a huge database. This method has been used to find the relationship between each set of items based on the interestingness measures such as support, confidence, lift, similarity measures, etc. By typical association rule technique, we generate association rule that satisfy minimum support and confidence values. Support and confidence are the most frequently used, but they have the drawback that they can not determine the direction of the association because they have always positive values. In this paper, we compared support, basic confidence, and three kinds of confidence measures useful for classification model building to overcome this problem. The result confirmed that the causal confirmed confidence was the best confidence in view of the association mining because it showed more precisely the direction of association.

• Journal of the Korean Data and Information Science Society
• /
• 제25권6호
• /
• pp.1345-1352
• /
• 2014

Data mining is the representative methodology to obtain meaningful information in the era of big data.By Wikipedia, association rule learning is a popular and well researched method for discovering interesting relationship between itemsets in large databases using association thresholds. It is intended to identify strong rules discovered in databases using different interestingness measures. Unlike general association rule, inverse association rule mining finds the rules that a special item does not occur if an item does not occur. If two types of association rule can be simultaneously considered, we can obtain the marketing information for some related products as well as the information of specific product marketing. In this paper, we propose a balanced attributable relative accuracy applicable to these association rule techniques, and then check the three conditions of interestingness measures by Piatetsky-Shapiro (1991). The comparative studies with rule accuracy, relative accuracy, attributable relative accuracy, and balanced attributable relative accuracy are shown by numerical example. The results show that balanced attributable relative accuracy is better than any other accuracy measures.