체육분야 패널데이터 구축 관련 연구자들의 주관적 인식 유형 연구

신상현,이완영 한국사회체육학회 2019 한국사회체육학회지 Vol.0 No.78

Purpose: The purpose of this study is to derive and analyze subjective types of subjective perception of panel data construction in physical education field. Method: Based on the existing literature research and in-depth interviews of the major engineers, the Q methodological typology was used to analyze the factors and correlations between subjective perceptions of panel data construction. For this purpose, 16 statements and 26 samples were used. Results: The most subjective perception type of panel data construction in the field of physical education was ‘Panel data construction indifference type’, followed by ‘Panel data construction requirement’ and ‘Panel data construction reconsideration’. This showed that there was still a lack of awareness in the field of physical education in terms of necessity and utilization unlike a lot of other academic fields that are collecting longitudinal data through various panel surveys and using them in various ways. First of all, it was necessary to expand the curriculum on data statistics to attract the attention of academics. There were many master’s and doctorate students each year, but few scholars major in physical education statistics. Therefore, it was necessary to establish professional education and institution for efficient management of sports data. Conclusion: In summary, it is shown to recognize the academic value of panel data and to make efforts to utilize panel data of other disciplines. Therefore, physical education is regarded as part of preparing for sports big data in the 4th industry that comes through the recognition that panel data construction is urgent. Accordingly, it is required to raise the awareness to raise the academic status by recognizing the necessity of longitudinal area data in addition to the existing cross-sectional data of many sports fields.

보건의료 연구를 위한 패널데이터: 리뷰 및 제언

박나영 ( Nayoung Park ),민상희 ( Sanghee Min ),우혜경 ( Hyekyung Woo ) 한국보건정보통계학회 2023 보건정보통계학회지 Vol.48 No.2

Objectives: Several panel surveys have recently been accumulated for various purposes in Korea, and vast amounts of data have been built over a long period of time. However, there are still no networks or platforms that can provide integrated information about existing panel data. In this paper, panel data, which is widely used on healthcare research, was reviewed and future suggestions were proposed in establishing panel data. Methods: This study classified and analyzed 9 types of panel data commonly used in healthcare research according to their status, sample composition, sample retention rate, survey items, and other characteristics. Results: The use of panel data is particularly useful for longitudinal analyses, such as examining health trajectories or identifying causal relationships with influencing factors. To utilize panel data effectively for healthcare research, efforts are needed to establish links between panel data and administrative data, provide post-survey management support, and ensure continuous quality improvement. Conclusions: It is very important to establish the integration platform for various panel data and administration data. If we develop strategies for quality management for it, the value of panel data and utilization of researchers will be enhanced.

Prediction of Bank Outstanding Customers Churn by Panel Data Size

김우중,안재준,오경주 계명대학교 자연과학연구소 2023 Quantitative Bio-Science Vol.42 No.2

Panel data is a type of data that combines cross-sectional and time-series information, providing a comprehensive dataset. Therefore, data collection, management, and analysis are difficult. To effectively use the data, the analysis methodology must be defined based on the size of the panel data. In this study, the performance of each prediction model was compared by analyzing the panel data to predict whether or not outstanding customers would churn based on the changes in the customer size and period. The algorithms used include a mixed model that incorporates random effects, primarily used in the panel data. Additionally, there are general statistical models and machine learning models that do not incorporate random effects. Based on the size of the panel data, panel, non-panel, linear-based, tree-based, and neural network-based models were compared. For each type, predictive performance rankings were derived. The analysis revealed that as the time period increased, the performance of the panel model improved in comparison to that of the nonpanel model. Additionally, the size of the customer did not have a significant impact. Furthermore, the performance of the base model was compared based on size. Based on the size of the panel data, it is expected that the appropriate model can be easily determined using these results. This study is significant because it establishes the effectiveness of the predictive models that can be applied to the panel data, which is typically challenging to analyze because of its size.

스포츠 패널데이터 구축 방안 연구

신상현,정호원 한국체육정책학회 2021 한국체육정책학회지 Vol.19 No.1

The purpose of This study examines the usefulness of panel data in terms of supplementing the deficiencies of sports-related statistics that are currently being generated, and attempts to reveal the demands of the times in establishing sports industry panel data in grasping, researching, and establishing policies in the sports. In this study, the opinions of experts were collected and analyzed, focusing on preparations and methods for building panel data in the sports. And based on the results of the expert group in-depth interview survey, the priorities for establishing sports panel data were derived. First of all, in order to prepare panel data, the method of converting the existing survey to the panel survey format should be considered first. Second, It was found that the subject of the panel survey needed to establish a research institute for a new data environment. Third, It should be a scientific investigation that can accurately estimate the probability of initial extraction of the survey sample or the probability of sample dropout occurring between waves. Lastly, The question of whether the panel survey period should be quarterly, half-yearly, one year or two years also depends on the purpose and content of the survey. As already pointed out, in consideration of the fact that the original data of the panel survey are disclosed to the outside, the method of publishing the original data and future research plans should also be considered in combination in selecting the panel survey cycle. Panel data construction is a long-term and continuous business, and the required budget is not small, so it is believed that it will help to reduce trial and error that may occur when building panel data.

스포츠 패널데이터(Panel Data) 구축을 위한 타당성 연구

신상현,정호원 한국체육정책학회 2018 한국체육정책학회지 Vol.16 No.2

The purpose of this study is to improve the information environment by selecting panels that are stable for samples for statistical analysis in various sports sectors. Therefore, based on the results of this study, we conducted a feasibility study on the status of domestic and overseas panel data and physical education statistics for the feasibility of deploying sports panel data, and conducted a feasibility study on the concrete development of panel data. Accordingly, there are four major possible consequences to be expected from deploying sports panel data. First, it is possible to secure data credibility by deploying panel data. Second, it is possible to lay the foundation for the expansion of research capabilities by deploying panel data. Third, it could lay the foundation for the establishment of a sports Big Data research center and the training of human resources. Finally, it is expected that panel data will be produced based on legal grounds and that physical education will be enhanced.

패널 데이터의 단위근 분석에 대한 새로운 접근

김윤대(Kim, Yundae),전치혁(Jun, Chi-Hyuck) 한국경영과학회 2010 한국경영과학회 학술대회논문집 Vol.2010 No.6

Panel data is a type of data that includes time-series and cross-sectional dimension. To analyze panel data, it should be known that it is stationary or non-stationary data. If the data is non-stationary and analyzed directly, it may lead to error. The panel unit root test determines if panel data is stationary or not. Many types of unit root test of panel data have been developed which included IPS unit root test and Fisher’s test. This paper presents a new panel unit root test using false discovery rate (FDR). After proposing the new model, this paper compares it with IPS and other models by some artificial data. It is concluded that the new model has similar power of test as compared with other tests.

패널 데이터의 단위근 분석에 대한 새로운 접근

김윤대(Kim, Yundae),전치혁(Jun, Chi-Hyuck) 대한산업공학회 2010 대한산업공학회 춘계학술대회논문집 Vol.2010 No.6

Panel data is a type of data that includes time-series and cross-sectional dimension. To analyze panel data, it should be known that it is stationary or non-stationary data. If the data is non-stationary and analyzed directly, it may lead to error. The panel unit root test determines if panel data is stationary or not. Many types of unit root test of panel data have been developed which included IPS unit root test and Fisher’s test. This paper presents a new panel unit root test using false discovery rate (FDR). After proposing the new model, this paper compares it with IPS and other models by some artificial data. It is concluded that the new model has similar power of test as compared with other tests.

SOLAR SEG 데이터세트 : 태양광 패널 표면 위 오염 세그멘테이션 데이터세트 구축

가을,최상현,나스리디노프 아지즈 한국콘텐츠학회 2024 한국콘텐츠학회논문지 Vol.24 No.7

최근 지구 온난화가 심화되면서 태양광 에너지의 수요가 급증하고 있다. 다만, 태양광 발전의 중요 부품인 패널은 실외에 설치되어 다양한 오염에 쉽게 노출된다. 이러한 오염은 발전량을 감소시키는 주된 원인이기 때문에 태양광 패널의 오염 여부를 조기에 감지할 필요가 있다. 오염을 정확하게 감지하기 위해서는 오염의 종류와 형태를 고려해야 한다. 하지만 이를 모두 고려한 데이터세트는 부족한 실정이다. 본 연구는 태양광 패널 위 오염의 종류 및 형태를 모두 고려한 SOLAR SEG 데이터세트(Soiling On soLar pAnel suRface for SEGmentation dataset)를 소개한다. 데이터 수집은 2023년 3월부터 5월까지 아침·점심·저녁 시간대에 진행됐다. 또한 효율적인 오염 생성을 위해 오염생성 사이클을 정의했다. 이 데이터세트는 오염된 태양광 패널 이미지와 그에 대한 세그멘테이션 라벨링 데이터를 포함한다. 또한 데이터세트의 유용성을 검증하기 위해서 이미지 세그멘테이션 모델인 FCN(Fully Convolutional Networks)를 이용한다. 해당 데이터세트로 훈련한 FCN은 정확도는 95.60%, 클래스 정확도는 90.78%, mIoU는 82.17%으로 높은 성능을 보였다. 이러한 실험 결과를 바탕으로 본 논문은 SOLAR SEG 데이터세트 구축과정을 구체적으로 기술함으로써 태양광 패널 표면 위 오염과 관련된 후속 연구에 기여한다. Due to the worsening global warming, the demand for solar energy is rapidly increasing. However, solar power generation has the problem of various soiling accumulations as the panels are exposed to the natural environment. To analyze such soiling, both the type and the shape of soiling must be considered. However, existing datasets often do not consider these factors. This study introduces the SOLAR SEG dataset (Soiling On soLar pAnel suRface for SEGmentation dataset), which considers both the type and shape of soiling on solar panels. Data collection was conducted during morning, afternoon, and dinner hours from March to May 2023. Additionally, a 'soiling generation cycle' was defined for efficient soiling generations. This dataset contains images of dirty panels and their segmentation labeling data. Additionally, Fully Convolutional Networks (FCN), an image segmentation model, is used to verify the usefulness of this dataset. The FCN trained with the dataset showed high performance with an accuracy of 95.60%, class accuracy of 90.78%, and mIoU of 82.17%. By describing the process of constructing this SOLAR SEG dataset, we contribute to subsequent research on soiling on solar panel surfaces.

데이터 마이닝 기법을 통한 교육 패널데이터 분석

유진은(兪鎭銀) 서울대학교 교육연구소 2016 아시아교육연구 Vol.17 No.3

명확한 기존 이론이 없어도 축적된 데이터 분석을 통하여 결과를 도출할 수 있는 데이터 마이닝 기법이 빅데이터 시대에 각광을 받고 있다. 수렴 또는 과적합 등의 문제로 인해 소수의 변수만을 모형화해 온 기존 연구방법과 달리, 데이터 마이닝 기법으로는 수백 개의 변수를 한 모형에 투입할 수 있으며, 따라서 연구방법적 측면에서 여러 장점을 가진다. 국가기관에서 십수년 간 수집해 온 교육 패널자료는 양적 · 질적인 측면에서 데이터 마이닝기법 적용에 적절하다. 본 연구는 빅데이터 분석에서 자주 이용되는 벌점회귀모형인 LASSO를 KYPS 5차 자료분석에 이용함으로써 데이터 마이닝 기법의 교육 패널자료 적용 사례를 제시하였다. 수십 개의 변수만을 이용하였던 기존 연구와 달리, 본 연구는 총 315개의 설명변수를 한 모형에 투입하여 15개의 변수를 선택하였다. 기존 연구에서 모형화된 변수뿐만 아니라 새로운 변수를 발굴할 수 있었다. 본 연구의 함의 및 후속 연구 주제 또한 논의되었다. With the advent of so-called big data era, data mining techniques have come to the fore as big data analysis tools. Unlike conventional statistical methods, data mining techniques can handle hundreds of variables in one model without convergence or overfitting problems. However, studies in the field of education have not yet paid enough attention to recent data mining techniques. Particularly, panel data with its hundreds of variables and thousands of participants can fit data mining techniques. This study aimed to illustrate a popular data mining technique, LASSO, by applying it to the 5th wave of KYPS (Korea Youth Panel Study). A penalized LASSO regression was executed with 10-fold cross-validation via deviance, and was successfully applied to the social sciences panel data. Implications of the study are discussed as well as further research topics.

대학교육 질 관리를 위한 학습자 패널 데이터 타당화 연구

이훈병 ( Lee Hunbyoung ) 아시아문화학술원 2017 인문사회 21 Vol.8 No.6

본 연구는 대학교육의 질 제고를 위한 패널조사의 주요영역 및 요소들을 선정하고 타당성을 제시하는데 목적을 둔다. 본 연구는 교수와 직원을 대상으로 내용타당도 비율검사, 요인분석, 신뢰도 분석을 실시하였다. 연구결과를 살펴보면 다음과 같다. 첫째, 대학교육 질 관리를 위한 학생 패널데이터는 입학, 학교생활, 졸업 및 취업 3영역에 총 34개(입학 9개, 학교생활 17개, 그리고 졸업 및 취업 8개) 항목이 패널데이터로서 타당성을 확보하였다. 둘째, 대학교육 질 관리를 위한 요인은 3영역 6개 요인으로 나타났다. 요인분석 결과 입학영역은 입학점수, 학과, 가족관계 요인으로, 학교생활 영역은 상담 및 진단검사, 학습활동성과와 기대, 전공, 진로 전망 요인으로, 그리고 졸업 및 취업 영역은 취업내용, 취업노력, 전공일치도 요인으로 분류되었다. 효과적인 패널 운영 및 패널데이터 수집 및 분석을 위해서는 첫째, 개인정보보호를 위한 대책이 요구된다. 둘째, 전담관리 조직 및 관련부서들의 협조가 필요하다. 셋째, 무응답 또는 표본이탈 방지 노력에 요구된다. 넷째, 패널자료의 적극적 이용 및 이용 활성화 방안이 모색되어야 한다. 추후 본 연구결과를 바탕으로 패널데이터를 이용한 다양한 교육성과 분석 연구를 모색하고자 한다. The purpose of this study is to select and validate the areas and items of the panel survey. The analytical methods used are the content validity ratio test, factor analysis, and reliability analysis. The results are as follows. Firstly, the panel data area for the quality management was selected as admission, school life, graduation&employment. A total of 34 items (9 for “entrance”, 17 for “school life”, and 8 for “graduation&employment”) were selected as the panel data. Secondly, the areas for panel data consist of three domains and six factors. In the factor analysis, the “entrance” consisted of entrance score, department, and family. The “school life” consisted of counseling&diagnostic tests, learning outcomes&expectations, major and career prospects. The “graduation&employment” consisted of the employment, efforts for employment, major adjustment. The suggestions are as follows. ① A plan for protecting personal information is required. ② It is necessary to create a management organization and cooperate with related departments. ③ It is required for the efforts to prevent non-response&sample deviation. ④ Utilization of panel data should be considered. Based on the results of this study, we will conduct various performance analysis studies using panel data.