클러스터 평가 외부기준 척도 Fn‐Measure

김경택 한국산업경영시스템학회 2012 한국산업경영시스템학회지 Vol.35 No.4

F-Measure is one of the external measures for evaluating the validity of clustering results. Though it has clear advantages over other widely used external measures such as Purity and Entropy, F-Measure has inherently been less sensitive than other validity measures. This insensitivity owes to the definition of F-Measure that counts only most influential portions. In this research, we present Fn-Measure, an external cluster evaluation measure based on F-Measure. Fn-Measure is so sensitive that it can detect their difference in the cases that F-Measure cannot detect the difference in clustering results. We compare Fn-Measure to F-Measure for a few clustering results and show which measure draws better result based upon homogeneity and completeness

클러스터 평가 외부기준 척도 F_n-Measure

김경태(Kyeongtaek Kim) 한국산업경영시스템학회 2012 한국산업경영시스템학회지 Vol.35 No.4

F-Measure is one of the external measures for evaluating the validity of clustering results. Though it has clear advantages over other widely used external measures such as Purity and Entropy, F-Measure has inherently been less sensitive than other validity measures. This insensitivity owes to the definition of F-Measure that counts only most influential portions. In this research, we present F_n-Measure, an external cluster evaluation measure based on F-Measure. F_n-Measure is so sensitive that it can detect their difference in the cases that F-Measure cannot detect the difference in clustering results. We compare F_n-Measure to F-Measure for a few clustering results and show which measure draws better result based upon homogeneity and completeness.

요양급여 적정성평가 지표관리 방안 - 지표별 목표 설정 및 모니터링을 중심으로 -

임지혜,윤여진,박소정,김정림 건강보험심사평가원 심사평가연구소 2020 연구보고서 Vol.2020 No.0

The focus of quality assessment has shifted from healthcare service use to clinical quality improvement. As of September 2018, the quality assessment in Korea examines 34 items with 408 measures. While quality assessment has expended in quality and quantity, the validity and confidence in assessment measures has decreased, affecting assessment efficiency negatively. In response, the Health Insurance Review & Assessment Service (HIRA), carried out a reshuffle of assessment measures from 2016 to 2019. Nevertheless, there is a lack of the measure lifecycle through measure development, implementation, maintaining, and retire. Against this backdrop, this study was designed to contemplate measure management based on effectiveness analysis of existing measures. For this study, 16 items were selected as target with measure that were continuously used 3 to 4 times in quality assessment as of 2019. The measurement type was analyzed for each measure, and distribution was observed based on descriptive statistics values, including average, standard deviation, applicable quartile, and minimum and maximum. Using AHRQ analysis of the U.S. and QOF analysis of the U.K., 95% performance achievement rate and average yearly fluctuation rate were analyzed. Results showed that there were ratio measures, case number measures, yes/no measures, and other measures. Among the 16 items, 79.49% were ratio measures, which was considered the most universal and suitable type to monitor achievement rate and average yearly fluctuation. As such, when AHRQ and QOF analysis were applied to ratio measures, most results were over 95% achievement rate, proving some level of quality improvement. Other measures were not suitable to use for performance achievement rate. Based on the analysis results, following suggestions were proposed for management of quality measures. Most measures does not come with a quality goal when measure development. To establish a measure lifecycle, there are five steps to follow: Step 1) the measure should be reviewed upon type of measurement, meaning of measures, and objective of measurement, Step 2) analysis method should be selected for the measure, Step 3) the results need to be analyzed, Step 4) validity of the measure should be reviewed, and Step 5) the type of measure management (maintaining, improvement, monitoring, retirement) should be decided. Also, for measures that are relatively new to assessment, the result of first assessment can be used as a benchmark for goal setting for next iteration. In conclusion, this study did not succeed in suggesting goal setting methods for existing measures, but it did propose a monitoring method based on assessment results. Form now on, newly developed measures should come with a goal and management frequency, and measure validation process needs to be added. To that end, a measure lifecycle should be put in place. Furthermore, a group of experts who contribute direction and thoughtful input to the measure development process requires to be formed to improve measure management and secure objectivity in operation.

A FORMATIVE MEASURE OF VERY YOUNG CHILDREN’S PERSUASION KNOWLEDGE

Shasha Wang,Dick Mizerski,Fang Liu,Doina Olaru,Victoria Mallinckrodt 글로벌지식마케팅경영학회 2016 Global Marketing Conference Vol.2016 No.7

Introduction An individual’s Persuasion Knowledge is his or her knowledge of the persuasive nature of messages (Freisted and Wright 1994). Most adults are assumed to have this knowledge that enables them to be sceptical of persuasive messages, and to incorporate this information in their decision-making about promoted products. Children are viewed as vulnerable to persuasive messages (Nelson 2016) because they have not developed the appropriate knowledge base to be sceptical of messages (Mallinckrodt and Mizerski 2007) and thus unable to discount the claims made by marketers. This lack of ability to be sceptical is argued to lead to “…the vulnerability of young audiences to the negative effects of advertising” (Nelson 2016, p. 169) like obesity, materialism, poor quality of life and higher rates of mortality. Because of the wide acceptance that young children are particularly vulnerable to persuasive messages, advertising targeting children has been banned in several countries and proposed for others (Mizerski et al. 2016; Wang 2016). Many studies about Persuasion Knowledge or Advertising Knowledge (communication research) have been published over the 40 plus years since Ward (1972) first discussed the concept. However, few studies have tested whether Persuasion Knowledge is an antecedent to children’s sceptical responses to persuasive messages. The 11 studies that have tested the link (see Mizerski et al. (2016) for a review) used a wide variety of single item or reflective measures (measures that reflect the construct). They provide inconsistent findings except that the child’s age is positively associated with acquiring knowledge about persuasive messages like advertising or playing advergames. For example, while most researchers assumed or argued a positive link between young children’s persuasion knowledge and their scepticism, Buijzen (2007) and Christenson (1982) failed to find this link. Robertson and Rossiter (1974) reported children’s understanding of persuasive intent (commercials persuade one to buy things) was positively related with young children’s scepticism, but assistive intent (commercials tell one about things) was negatively related. The inconsistent findings of children’s responses to commercial messages may be due to more than the lack of consistent measures. The use of reflective rather than formative measure of young children’s persuasion knowledge may be another reason for inconsistent findings. “Young” children are those under eight years old (Mizerski 1) shashaatperth@gmail.com et al. 2016), but a lack of sufficient persuasion knowledge has been found with children over 16 years old (Carter et al. 2011). Most recent studies have adopted the Freisted and Wright (1994) Persuasion Knowledge Model that is based on the information processing of an adult buyer. Adults tend to have obtained Persuasion Knowledge so their knowledge may be captured with measures that reflect the construct. Young children are in the process of obtaining Persuasion Knowledge. The ability to understand the source of the message and the persuasive intent of the source are often cited as antecedents to having Persuasion Knowledge. These constructs form over time and cause persuasion knowledge (Nelson 2016). Therefore, it should be a formative measure. Measuring social class is a classical formative measure because causal elements like where you live and your profession can’t be accurately calculated for children. Social class forms over time. The use of a reflective measure when a formative measure should be used leads to several problems (Diamantopoulos and Siguaw 2006; Diamantopoulos and Winklhofer 2001), particularly an increase in Type II errors-“false negatives” (MacKenzie, Podsakoff and Jarvis 2005). For example, this means ruling out a causal element of persuasion knowledge when it is causal. This paper will compare existing single item or multi-item reflective measures used with young children, with a formative measure of the Persuasive Knowledge construct. The best way to validate a construct is to test it with external variables empirically and theoretically linked to the construct, including both antecedents and consequences (Diamantopoulos and Winklhofer 2001). Three variables, theoretically and/or empirically linked with young children’s Persuasion Knowledge, are tested in a baseline model to assess the external validity of the construct. These variables are age (Ward 1972), responses toward persuasive marketing messages (such as scepticism) and affect toward the persuasive messages (Mizerski et al. 2016; Wang 2016). Therefore, it is expected that the goodness-of-fit measures for the model using the formative measure of young children’s Persuasion Knowledge will provide a better fit to the data than the reflective measures. To further test the formative nature of young children’s Persuasion Knowledge, two additional models are tested. Researchers are responsible to set the weights of indicators of a formative construct, so a formative model with expert knowledge weights is developed (Figure 1). The indicators or elements of a formative construct should be able to reveal different facets so another model with different facets is developed. Consistent findings of the two models and the proposed baseline model will further support the formative nature of this construct. Apart from content validity and external validity, we also test the measures of the construct with another data set (Mallinckrodt and Mizerski 2007) to test the models’ generalisability. The Mallinckrodt and Mizerski study used children from a different cultural background (Australian vs. Chinese young children), but have similar ages and measures of Persuasion Knowledge and external variables. Further confirmation of the structure of the measurement model is provided if the same relationships are found with the second data set. Methodology Sample The population to be sampled are young Chinese children. China was selected because it has the largest population of young children, is the largest market for toys and a children’s toy is the stimulus product in the experimental study. The sample frame is day care schools in a Northern Chinese city of approximately eight million people. Procedure This is an experiment-based study with a control group. After individual exposure to a toy TV advertisement for a “magic ruler” that can be made into many different shapes with a Dinosaur shape shown in the ad, participants were each asked to answer questions. Cartoon pictures of the question options accompanied by verbal statements were used to reduce the possibility of misunderstanding young children’s responses on Persuasion Knowledge related questions. To reduce any effects of young children choosing the first option they see, pictures or options were shown in a random order. Children were told that there is no right or wrong answer, and they could withdraw at any point. Measures Persuasion knowledge was measured three ways; including a single-item measure, a summated-items measure, and a formative measure. Through an analysis and coding of 20 studies that tested the effect of young children’s persuasion knowledge (Mizerski et al. 2016), six items were found to measure the Persuasion Knowledge construct. The single-item measure used is the children’s understanding of the advertisers’ intention to make them ask their parents to buy (parent-buying intent). This was treated as the most important aspect of persuasion knowledge by several researchers (Carter et al. 2011; Mallinckrodt and Mizerski 2007). A summated-items measure included six items, frequently used in prior scholarly work. The formative measurement model was built using the same six items but by changing the direction of influence, with the causal flow from measures to the construct. Other variables include the children’s scepticism, their belief of false claims made in the ad and affect toward the toy TV advertisement (see Figure). After a CFA analysis with five questions, the scepticism factor score was derived as a standardised measure that followed a normal distribution. Scepticism ranged from -0.59 to 1.35, with an average of 0.07 and SD of 0.68. Belief of false claims (named as “false beliefs”) shown in the TV advertisement indicated that most children did not believe the two false claims included in the ad (84% and 69% respectively). Affect towards the toy TV ad was measured using one question: “Do you like this video?” to which most children (88%) indicated yes. Results, Discussion And Implication Fifty-four different patterns or combinations of the six persuasion knowledge items were found. This pattern show substantial heterogeneity in children’s Persuasion Knowledge, and further illustrates that young children are accumulating or forming their Persuasion Knowledge (Friestad and Wright 1994). In addition, these items have low correlations to one another that are typical of a formative measure. These findings of variability of knowledge levels and weak association between them indicate the potential multiple-dimension, formative nature of the construct of persuasion knowledge for young children. This may apply specifically to young children who are at the stage of increasing their learning abilities and developing or forming their knowledge. No relationship was found in the structural models using the reflective single-item measure and summated-items measure of persuasion knowledge, and the models showed a poor fit. However, relationships were found in the structural model that applied the formative measure of persuasion knowledge, with good model fit (see Table 1). Because a formative measure is supported, we tested the formative nature of Persuasion Knowledge on a previous study’s data (Mallinckrodt and Mizerski 2007). A formative model using that study’s data showed the same relationships between Persuasion Knowledge and its external variables. This consistency supports a formative measure of young children’s persuasion knowledge. Any review of future or past research should note the possible impact of using reflective measures of young children’s Persuasive Knowledge. To generalise the findings more research needs to be done for different product categories and age groups. More product categories, such as food and movies, and age groups could be taken into consideration. While few studies have tested the association of young children’s Persuasion Knowledge to scepticism toward the message, even fewer have tested the link of scepticism to young children’s responses to the advertised product (e.g. like, prefer, choose). Most of these studies do not find a link. If having Persuasion Knowledge doesn’t influence a young child’s desire for the brand, why teach it (e.g. Nelson 2016) or ban advertising because the children don’t have Persuasion Knowledge? Perhaps using a formative measure the link will be found.

중국의 섬유무역에 적용되는 특별 세이프가드조치 및 시사점

최송자 법무부 2011 통상법률 Vol.- No.101

세이프가드조치는 공정무역에 적용되는 무역규제수단으로, 일반 세이프가드조치와 특별 세이프가드조치로 분류된다. 특별 세이프가드조치의 무역규제효과는 일반 세이프가드조치보다 훨씬 크다. 20세기 1990년대 중반부터 중국은 세계 최대의 섬유 생산국 및 수출국으로 급부상하면서 세계 섬유무역에서 막강한 영향력을 행사하고 있다. 2001년 말 중국이 WTO에 가입한 후 중국 섬유무역의 발전을 견제하는 중요한 수단으로 특별 세이프가드조치가 적용되고 있었다. 이는 중국이 WTO 회원국으로서 기타 회원국과 동등하게 적용받는 특별 세이프가드조치와 중국의 WTO 가입양허에 따라 중국산 섬유수출에만 적용되는 특별 세이프가드조치가 포함된다. 중국의 WTO 가입양허에 따라 중국산 섬유수출에만 적용되는 특별 세이프가드조치는 2013년 12월 11일에 효력을 상실하게 된다. 따라서 2005년 1월 1일부터 세계 섬유무역은 일반 세이프가드조치의 적용을 받고 있지만, 중국의 섬유수출만은 현재까지도 여전히 특별 세이프가드조치의 적용을 받고 있다. 그러나 이런 상황은 2013년 12월 11일 이후 크게 변화될 것이다. 對중국 특별 세이프가드조치가 효력을 상실한 후 세계 섬유무역에서 중국의 영향력은 더욱 막강해 질 것이다. 이 논문에서는 먼저 세계 섬유무역과 중국 섬유무역이 특별 세이프가드조치와 어떤 관련이 있는지 각각 살펴보고 있다. 그 다음 중국의 섬유무역에 적용되는 특별 세이프가드조치 간 비교를 진행하고 있다. 마지막으로 對중국 특별 세이프가드조치가 효력을 상실한 후 세계 섬유무역에 일으킬 영향력 및 우리 섬유산업의 대응방향에 대해서 제시하고 있다. Safeguard measure is a way to control trade applied to fair trade. It is classified as either general safeguard measure or special safeguard measure. The effect of the special safeguard measure is greater than that of the general safeguard measure. In the 20th century, China became the world's largest producer and exporter of textile and has enjoyed an enormous influence in the world textile trade since the mid 1990s. Since China joined the WTO (World Trade Organization) in 2001, the special safeguard measure has been applied as a way to hold the advancement of Chinese textile trade. This measure includes not only the general one applied as a member of WTO, the same way as every other member of the organization, but also a separate special safeguard measure which is applied to the Chinese textile products manufactured for export purposes, due to the concessional term made upon joining the WTO. This separate special safeguard measure is going to be expired December 11, 2013. Therefore, the general safeguard measure is applied to the world textile trade while the special safeguard measure is still being applied only to China since January 1, 2005. This situation, however, will be changed when the special safeguard measure expires in December 11, 2013. When the special safeguard measure expires, it is expected that China will have a greater influence in the world textile trade. In this paper, the relations of the special safeguard measure with both the world and Chinese textile trade are analyzed. Next, the special safeguard measure of Chinese textile trade is evaluated in a comparative manner. At last, this paper discusses the potential influences on the world textile trade when Chinese special safeguard measure expires and suggests the ways to deal with the influences for Korean textile industry.

육계 계열화사업의 사육평가 유형에 따른 생산성 분석

강병규 ( Byung Kyu Kang ),안상돈 ( Sang Don Ahn ),김정주 ( Jung Joo Kim ) 한국축산경영학회 2012 농업경영정책연구 Vol.39 No.1

This study aimed to contribute to devising approaches for developing a broiler integration system and a method for evaluation by examining evaluation methods using empirical analysis and conducting an analysis and management efficiency and productivity. Overall efficiency in the CCR model was higher in the order of relative performance measure, absolute performance measure and portion relative performance measure. Technical efficiency in the BCC model was higher in the order of relative performance measure, absolute performance measure and portion relative performance measure. Scale efficiency was higher in the order of absolute performance measure, relative performance measure and portion relative performance measure. In accordance with the analysis results on the yearly change of the Malmquist Productivity Index (MPI), a relatively effective frontier change was caused by the improvement of management efficiency rather than technical production change in all three performance measures. It could be concluded that feeding and management by farms had a more significant impact on productivity than the improvement of a Feed Conversion Ration (FCR) by integration companies. As a result, the establishment of broiler integration system performance measure indexes, which are unique for Korea, is needed. a change from the present relative performance measure to performance measures which eliminate the top and bottom 10% is required.

연관성 평가를 위한 기여 순수 부호 헬링거 측도의 제안

박희창(Hee-Chang Park) 한국자료분석학회 2020 Journal of the Korean Data Analysis Society Vol.22 No.3

데이터의 중요성이 점차 확대되어 가고 있는 동시에, 이를 활용하기 위한 시도가 가파르게 증가하는 시대에 의미 있는 정보를 선별해내기 위해 많은 기법들이 개발되고 있다. 본 논문에서는 연관성 평가 기준 중에서도 기존의 부호 헬링거 측도와 순수 부호 헬링거 측도에 비해 연관성 규칙 평가 기준으로 좀 더 적합한 기여 순수 부호 헬링거 측도를 제안하였으며, 예제를 통하여 동시 발생 빈도와 불일치 빈도의 증감에 따른 변화 양상을 파악함으로써 측도의 유용성을 알아보았다. 기여 순수 부호 헬링거 측도를 기존의 부호 헬링거 측도와 비교해보면 절대값의 크기가 부호 헬링거 측도에 비해 기여 순수 부호 헬링거 측도가 더 큰 값으로 나타났으며, 각 사례들 간의 값들을 비교해보면 부호 헬링거 측도보다는 기여 순수 부호 헬링거 측도가 변화량이 더 큰 것을 알 수 있다. 또한 순수 부호 헬링거 측도와 기여 순수 부호 헬링거 측도의 값을 비교해보면 이 경우에도 부호 헬링거 측도의 결과와 유사하게 나타났다. 따라서 본 논문에서 제안하는 기여 순수 부호 헬링거 측도가 연관성 생성 기준으로서는 기존 측도인 부호 헬링거 측도 또는 순수 부호 헬링거 측도보다는 더 유용하다고 할 수 있다. In the age of increasing importance of data and attempts to utilize it, various techniques are being developed to select meaningful information. In this paper, among the association evaluation criteria, we proposed a attributable pure signed Hellinger measure that is more suitable as a association rule evaluation criterion than the existing signed Hellinger measure and pure signed Hellinger measure. The usefulness of the measure was investigated by grasping the change pattern according to. When comparing the attributable pure signed Hellinger measure with the existing signed Hellinger measure, it was found that the magnitude of the absolute value has a larger value of the attributable pure signed Hellinger measure than the signed Hellinger measure. By comparison, it can be seen that the attributable pure signed Hellinger measure has a greater change than the signed Hellinger measure. Therefore, it can be said that the attributable pure signed Hellinger measure proposed in this paper is more preferable than the existing measure, the signed Hellinger measure and the pure signed Hellinger measure.

판례연구 : 일반 행정법상 당초처분과 감액처분의 법리 검토 -대법원 2012. 9. 27. 선고 2011두27247 판결을 중심으로-

류광해 ( Kwang Hae Ryu ) 홍익대학교 법학연구소 2013 홍익법학 Vol.14 No.3

검토 대상판결인 대법원 2012. 9. 27. 선고 2011두27247 판결은, 부당이득금 환수와 관련하여 당초처분을 감액하는 처분이 이루어진 경우 징수의무자가 다투어야 하는 소송의 대상은 ‘감액처분에 의해 감액되지 않고 남은 당초처분’이고, 제소기간도 당초처분이 기준이 되어야 한다고 판시하였다. 이러한 대상판결의 법리는 조세쟁송에서 감액경정처분에 대한 기존의 대법원 법리 및 기존의 과징금 부과처분이나 개발부담금 부과처분에서 밝힌 법리와도 동일한 법리이다. 그런데 이러한 대상판결의 법리는 처분서의 문언 내용에 따라 처분의 내용을 판단하여 야 한다는 대법원의 판례 자체에도 부합하지 아니하고, 감액처분시 행하는 새로운 납부기일과 불복고지로 갖게 되는 징수의무자의 신뢰에도 반하며, 그 법리가 이해하기 어려워 국민의 권익보호에 기여하는 법리로 평가되기도 힘들고, 소송요건 중 제소기간의 측면에서 볼 때 징수의무자의 본안판단의 기회를 보호하는 법리로 보기 어려우며, 감액처분을 통하여 행정청 스스로 위법하다고 평가한 당초처분의 처분사유를 가지고 당초처분의 적법성을 판단하는 것은 잘못이라는 점 등의 문제점이 있다고 평가할 수 있다. 반면 처분서의 문언에 충실하여 감액처분 처분서의 문언이 ‘일부 취소’임이 명백하지 아니한 이상 ‘새로운 처분’으로 보아야 한다고 보면, 위와 같은 문제점은 없어지거나 크게 완화된다. 그러나 이 견해에도 처분변경으로 인한 소의 변경이 이루어지지지 아니함으로서 당초처분에 대한 소가 각하될 위험성이 있다는 지적이나 당초처분에 기초한 압류처분 둥후속처분의 효력이 문제될 수 있다는 지적은 가능하다. 그러나 전체적으로 볼 때 대상판결의 법리보다는 처분서의 문언에 충실하게 해석하는 것이 처분의 해석의 측면에서나 국민의 신뢰와 권익보호의 측면에서 올바른 견해라고 생각된다. In Sep. 27. 2012, in case of amount obtained by unjust enrichment, the Supreme Court sentenced that the object of litigation between a beginning administrative measure and a follow-up reduced administrative measure is a beginning measure reduced by the follow-up measure, and that the starting point of period of litigation is a beginning administrative measure. The above precedent principle coincides with the principle that the Supreme Court sentenced in the case of tax, penalty and development charge litigation. But the above principle does not coincide with the Supreme Court`s another principle that the meaning of a administrative measure should be interpreted by the contents of administrative measure documents. And the above principle betrays the responsible person`s trust that is made by disposition agency who gives the person another deadline for payment or notifies the objection methods when he disposes the following-up reduced administrative measure. And the above principle is too difficult to be understanded by public people, so it does not contribute to the safeguard of the public`s rights. And in terms of period of litigation, the above principle does not protect the responsible person`s right of procedure on the merits of a case. And it is wrong to judge the beginning administrative measure`s legality by the resons that is confirmed illegal through the follow-up reduced administrative measure. So I think it is right that the object of litigation between a beginning administrative measure and a follow-up reduced administrative measure is a follow-up reduced administrative measure, and that the starting point of period of litigation is the follow-up reduced one.

참여적 목표설정이 성과평가지표 탈락에 미치는 영향

김범준(Bum-Joon Kim) 한국경영학회 2020 經營學硏究 Vol.49 No.2

This study investigates how the ratee’s participation in the target setting process affects performance measure drop. In general, we evaluate performance by comparing actual performance to target in quantitative performance measures. It is very interesting research topic to examine whether the level of ratee’s participation in target setting process affects the performance measure selection because it is different by target setting method of performance measure. According to the prior literature, it is known that participative target setting provides the ratees the perception of procedural fairness in the evaluation process, thereby motivating them to work hard. In the other side, there is also an incentive for ratees to utilize private information to make budgetary slack in participative target setting process when there is information asymmetry between rater and ratee. Thus, I expect that designers of a performance evaluation system will consider countervailing two incentives when he decides to drop performance measure. Using performance evaluation results of Korean public companies from 1985 to 2011, I investigate how participative target setting affect performance measure drop. As a result, I find that participative target setting does not significantly affect the performance measure drop. However, I find that the common measure is relatively less likely to be dropped compared to the unique measure because common measure is more informative than unique measure due to relative performance evaluation. These findings suggest that participative target setting influences performance measure drop differently depending on the performance measure characteristics like common versus unique measure. This study contributes to expand research stream in performance measure choice.

순수 수정된 대칭적 J 측도에 의한 연관 규칙의 생성

박희창 한국자료분석학회 2019 Journal of the Korean Data Analysis Society Vol.21 No.4

Today, various data mining techniques and methods have been developed to analyze huge amounts of data in the context of big data (Park, 2017a). Data mining is a technique for finding meaningful information that is latent in a large database. In this paper, we proposed a purely modified symmetric J measure that transformed J measure among the thresholds used in the association rule technique. In addition, we used the example to compare various measures related to J measure - symmetric J measure, modified symmetric J measure, and purely modified symmetric J measure. As a result of comparing several J based measures with increasing coincidence frequency, inconsistency frequency, and frequency of incidence of non-occurrence, J measure and symmetric J measure in all cases were always decreased and then changed into the form of increasing curve, while modified symmetric J measure and purely modified symmetric J measure increased. The variation of purely modified symmetric J measure was larger than that of modified symmetric J measure. Therefore, it was considered that purely modified symmetric J measure was the most desirable as association evaluation criterion. 오늘날 빅 데이터의 환경 하에서 엄청난 양의 데이터를 분석하기 위해 보다 고급적이면서 다양한 데이터 마이닝 기법이 개발되었다(Park, 2017a). 데이터 마이닝은 방대한 빅 데이터베이스에 잠재되어 있는 의미 있는 정보를 탐색하는 기술이다. 본 논문에서는 연관성 규칙 기법에서 이용되는 평가 측도들 중에서 J 측도를 변형한 순수 수정된 대칭적 J 측도를 제안하였으며, 예제를 이용하여 J 측도와 관련된 여러 형태의 측도들인 대칭적 J 측도, 수정된 대칭적 J 측도와 비교하였다. 동시 발생 빈도, 불일치 빈도, 그리고 동시 비발생 빈도의 증가에 따라 여러 측도들을 비교해본 결과, 모든 경우에 있어서 J 측도와 대칭적 J 측도는 항상 양의 값만을 취하면서 감소하다가 다시 증가하는 이차 곡선의 형태로 변하는 반면에 수정된 대칭적 J 측도와 순수 수정된 대칭적 J 측도는 양과 음의 값을 동시에 취하면서 증가하였다. 또한 수정된 대칭적 J 측도와 순수 수정된 대칭적 J 측도 중에서는 순수 수정된 대칭적 J 측도의 변화 폭이 수정된 대칭적 J 측도보다 더 크게 나타나서 연관성 여부를 좀 더 명확하게 파악할 수 있었다. 따라서 순수 수정된 대칭적 J 측도가 본 논문에서 고려하는 측도들 중에는 연관성 평가 기준으로서 가장 바람직한 것으로 판단된다.