지적장애학생의 지각-운동능력에 의한 장애등급 판별의 예측요인 탐색

윤지운(Jiwun Yoon),한민규(Min Kyu Han) 한국발육발달학회 2016 한국발육발달학회지 Vol.24 No.1

The purpose of this study was to application of a discriminant function analysis based on perceptual and motor ability that are sub factors as body perception, space perception, direction perception, and time perception for students with intellectual disabilities. For the purpose, 238 students (161 male students and 77 female students) with intellectual disabilities ware recruited from Seoul, Kyunggi and Kyungbook at three site in Korea. The test of perceptual and motor abilities for students with intellectual disabilities, developed by Yoon (2015) was applied to measure perceptual and motor abilities for this study. Means and standard deviations were calculated and compared to give a general description about the perceptual and motor abilities of students with intellectual disabilities, and discriminant function analysis in IBM SPSS 21.0 was used. As a result, correct classification rate was 59% in male and 57.1% in female for levels of students with intellectual disabilities for discriminant analysis. The result shows that various educational approaches based on levels of intellectual disabilities are needed for developing perceptual and motor abilities of students with intellectual disabilities.

청각장애와 비장애 학생의 신체활동 수준에 따른 건강체력 비교

윤지운 ( Jiwun Yoon ),한민규 ( Minkyu Han ) 한국체육대학교 체육과학연구소 2014 스포츠사이언스 Vol.31 No.2

The object of this study is to analyse and compare health-related physical fitness of hearing-impaired and non-disability students according to their physical activity level· This research was performed on two groups of 80 students(40 male and 40 female) each, one for hearing impaired group, whose members were attending at a hearing disability school in the metropolitan area, and the other for non-disability group, whose students were recruited from B middle school in Gyunggi-do· To estimate frequency of physical activity participation, method of Goldin Leisure-Time Exercise Questionnaire(GLTEQ) was introduced, and cardiorespiratory endurance(PACER), muscular strength(grip), and flexibility(sit and reach test) were selected as health fitness test items from Physical Activity Promotion System(PAPS) list· Collected data was processed by SPSS 18·0 statistics package, and data analysis included mean and standard deviation of each items· Two-way ANOVA was performed for physical activity level according to gender difference and disability presence, and verification of average difference of health fitness items between groups, and also, for examining correlation effect· The results of the study showed that hearing- impaired students`` physical activity participation level was higher than non-disabled students(p<·001) and, however, hearing-impaired students`` health-related physical fitness level was lower then non-disabled students(p<·001)· The results indicate that health-related physical fitness decline was caused by inappropriate physical activity program rather than physical activity participation level·

PageRank 알고리즘을 활용한 체조도마 선수 역대 랭킹

윤지운 ( Jiwun Yoon ),박재현 ( Jea-hyeon Park ) 한국스포츠과학원 2017 체육과학연구 Vol.28 No.2

[Purpose] The purpose of this study is to calculate the ranking of vault players in artistic gymnastics by individual and by country using the PageRank algorithm. The purpose of this study is to provide basic data that can be used in gymnastics events by comparing the performances of historical vault players with those of previous Olympic competitions. [Methods] The data collected for this purpose is a score of 117 vault players based on the results of the Olympic final event published in the International Gymnastics Federation (FIG). For data analysis, PegeRnak algorithm was used for calculating the ranking of vault players using MS-Excel and NetMiner. [Results] The results are as follows that. First, the PageRank algorithm is possible to calculate for historical vault players` rankings. Specifically, the ranking of vault players for historical Olympic calculated by PageRank ranked as Gervasio Deferr from Spain (ESP) at 1st, Alexei Nemov from Russia (RUS) at 2nd, and Klaus Koste from Germany (GDR) at 3rd. Second, Network of vault players` ranking is separated by each generation of Olympic games. [Conclusions] As a conclusions, it is possible to calculate national ranking of vault games of gymnastics by using PageRank algorithm.

지식네트워크 분석을 활용한 특수체육의 연구동향 분석

윤지운 ( Jiwun Yoon ) 한국특수체육학회 2017 한국특수체육학회지 Vol.25 No.1

이 연구는 지식네트워크 분석을 활용하여 특수체육분야의 연구동향을 시계열적으로 분석함으로써 특수체육의 발전에 의미 있는 자료를 제공하는데 목적이 있다. 이때 지식네트워크 분석 방법 중에서 키워드 네트워크를 적용하였다. 이 목적을 달성하기 위하여 활용된 자료원은 2000년부터 2014년까지 15년 간한국특수체육학회지에 게제 된 논문 488편이며 1840개 키워드이다. 연구의 흐름을 구간별로 확인하기 위하여 5년 단위로 구분하여 분류하였으며(제1구간: 2000년-2004년, 제2구간: 2005년-2009년, 제3구간: 2010년-2014년) Excel 프로그램과 NetMiner 소프트웨어 프로그램을 활용하여 빈도분석, 중앙성분석 및 변화율 분석을 실시하였다. 연구결과는 다음과 같다. 첫째, 한국특수체육학회지에서 과거 15년 동안 6회이상 출현된 키워드는 37개이며 지적장애(92회), 지체장애(33회), 통합체육(22회) 등 소수 키워드의 독점화 현상을 확인할 수 있었다. 둘째, 연결중앙성을 중심으로 본 상위10 키워드는 제1구간, 제2구간, 제3구간으로 범주화된 구간별 차이가 있다. 셋째, 한국특수체육학회지에 출현하는 중심키워드의 변화는 제1구간을 기준으로 제2구간 그리고 제1구간을 기준으로 제3구간 간 상승키워드와 하강키워드에는 차이가 있다. 이 연구의 분석결과는 특수체육의 연구동향, 타 학문과의 연계, 신규 연구주제를 선정하는데 유용한 정보로 활용 할 수 있다. The purpose of this study was to provide the trends of research in adapted physical activity area using knowledge network analytic method. For this purpose, 1840 keywords were extracted from 488 research papers in the Journal of Adapted Physical Activity & Exercise (JAPAE) during 15 years from 2000 to 2014. In order to identify the trends, the data were divided by three terms into every 5 years (1st term from 2000 to 2004, 2nd term from 2005 to from 2009, 3rd term from 2010 to 2014). Frequency, centrality in networks and rate of change were applied using by Excel and Netmnier software. The results are as follows: first, there were 37 keywords that appeared more than 6 times in the past 15 years in the JAPAE, and it was the phenomenon of monopolization of majority keywords such as intellectual disability (92 times), physical disability (33 times), and integrated physical education (22 times). Second, There was a difference in the categorized Top 10 keywords in each term of between centrality (compare with first to second term, and compare with first and third term). Third, major keywords appearing in the JAPAE have changed over 15 years. The results of this study can be used as a useful information for research trends of adapted physical activity area.

스포츠 AI 심판에 대한 언론의 인식: 키워드 네트워크 분석 활용

김세화(SeHwa Kim),윤지운(Jiwun Yoon),박재현(ae-Hyeon Park) 한국체육측정평가학회 2024 한국체육측정평가학회지 Vol.26 No.4

The purpose of this study is to analyze the media’s perception of sports AI referees through keyword network analysis. To achieve the purpose of this study, a total of 4,832 news articles related to AI referees were selected as research data from Bigkinds, the media database of the Korea Press Foundation. For keyword network analysis, data preprocessing was performed using the KrKwic program. Afterwards, 1-mode network analysis was performed using Netminer 4.0 and the research results were visualized. The conclusions of this study are as follows. First, the perception of AI umpires generally showed high frequency and connectivity with baseball-related keywords, which confirmed that while there were positive evaluations of the accuracy and fairness of AI umpires, there were negative evaluations of machine defects and errors. did. Second, the perception of AI referees by event was confirmed in three events. In baseball, controversy was confirmed regarding the issue of introducing the Automatic Pitching System (ABS) and the potential problems of such AI umpires.In soccer, the use of AI referees and technical errors were confirmed. In gymnastics, expectations for the introduction of AI judges in the Olympics and an accurate and fair scoring system were confirmed. This study confirms through news articles in media reports that while there are positive expectations about the accuracy and fairness of AI referees in sports, there are still concerns about AI referee errors, misjudgments, and doubts about overall reliability issues. These results provide a new perspective to evaluate the overall capabilities of AI referee systems. In follow-up research, I plan to analyze various sports events and situations and use them as basic data to find ways to continuously develop and improve the AI referee system.

유럽 5대 리그에서 UEFA 챔피언스리그를 진출하기 위한 최적의 승점은?

박경훈 ( Kyung-hoon Park ),김대건 ( Daegeon Kim ),윤지운 ( Jiwun Yoon ) 한국체육대학교 체육과학연구소 2023 스포츠사이언스 Vol.41 No.3

This study aimed to find out the optimal points for each European football league to advance to the UEFA Champions League in European professional football. Data were collected on whether or not the top 5 European leagues participated in the UEFA Champions League for a total of 22 seasons from the 2000~2001 season to the 2021~2022 season, the points in the first half and final points of the league. One-way ANOVA and ROC analysis were performed. ROC analysis calculated the Area Under the Curve (AUC), and the optimal cut-point by Youden’s J index. Sensitivity, Specificity, Accuracy, Balanced Classification Rate (BCR), and F1-score (harmonic mean of recall and precision) values were calculated to verify the validity of the optimal cut-point. Our results are as follows: In the first half of the season, Premier League, LaLiga, and Ligue 1 scored the highest with 31.5 points, followed by Italy with 30.5 points and Bundesliga with 26.5 points. In terms of final points, the Premier League had the highest score of 65.5 points, Serie A 64.5 points, Ligue 61.5 points, La Liga 60.5 points, and Bundesliga 54.5 points.

딥러닝 기법을 활용한 투척경기 성과 예측 가능성 탐색

이영선 ( Young Sun Lee ),박재현 ( Jaehyeon Park ),윤지운 ( Jiwun Yoon ) 한국체육대학교 체육과학연구소 2024 스포츠사이언스 Vol.42 No.3

The purpose of this study was to develop a performance prediction model for throwing event athletes using Deep Learning, XGBoost, and an ensemble model, thereby enhancing the efficiency of athlete training and improving performance. The methodology involved constructing Deep Learning, XGBoost, and ensemble models based on various basic and specialized physical fitness variables. The data preprocessing included standardization, and the performance of each model was evaluated using cross-validation techniques. The results showed that the Deep Learning model achieved the highest prediction accuracy with an R-squared value of 0.72, indicating its superior performance. The XGBoost model also demonstrated predictive capability with an R-squared value of 0.29. The ensemble model, which combined the predictions of both models, recorded an R-squared value of 0.49, indicating relatively better performance. These findings suggest that the combination of Deep Learning and tree-based models can effectively capture the nonlinear characteristics and interactions of variables in throwing events, although the Deep Learning model exhibited the most superior performance. However, this study has limitations in terms of the number of cases and the scope of the data. Thus, future research should involve more comprehensive studies with diverse variables and datasets to deepen the understanding.

토픽모델링을 적용한 스포츠 승부조작의 판례분석

이지용(Jiyong Lee),박재현(Jaehyeon Park),윤지운(Jiwun Yoon),이승박 (Seungbak Lee),염세경(Sekyoung Youm),이영섭(Yungseop Lee),윤효준(Hyojun Yun) 한국체육측정평가학회 2021 한국체육측정평가학회지 Vol.23 No.2

이 연구는 토픽모델링을 적용한 판례분석을 통해 승부조작의 주요 토픽을 확인하였다. 이를 위해 승부조작 주제로 진행된 판례 사례를 수집하였으며, 대한민국법원(http://www.scourt.go.kr)을 자료원으로 선정하였다. 자료수집 절차는 첫째, “승부조작”을 검색의 키워드로 선정하였다. 둘째, 판결선고 날짜 기준 2010년부터 2019년까지 53건의 판례를 선정하여 원심과 항소로 구분하였다. 셋째, 원심과 항소가 같은 사건을 다루는 경우 하나의 문서로 통합하여 최종적으로 43건의 판례가 분석자료로 선정되었다. 넷째, 판례를 시기와 종목에 따라 분류하였다. 토픽모델링은 LDA 알고리즘을 적용하였으며 모든 자료처리는 python 3을 사용하였다. 토픽모델링 결과와 토픽별 주요 판례 원문을 검토한 결과, ‘승부조작 경로’, ‘E-Sports와 관련된 승부조작’, ‘전통적 승부조작’, ‘승부조작을 식별하는 방법’, ‘양형을 판단하는 근거’가 판례에 나타나는 승부조작의 이슈임을 확인할 수 있었다. 이러한 결과는 향후 승부조작과 관련된 다양한 정책 개발 및 교육의 기초 자료로 활용이 가능할 것이다. The present study identified the major topics for match fixing by analyzing precedents through the use of topic modeling. Toward this purpose, the study collected cases of precedents on the topic of match fixing and chose the Court of Korea(http://www.scourt.go.kr) as a data source. First, the procedures for data collection involved searching the keyword “match fixing”. Second, the study selected precedents from 53 cases from 2010 to 2019 based on the pronouncement of judgment date and divided them into original trials and appeals. Third, if an original trial and appeal dealt with the same case, they were combined into a single document so that ultimately precedents from 43 cases could be selected as analysis data. Fourth, these precedents were categorized according to period and event. Topic modeling applied the LDA algorithm, and all data processing made use of Python 3. The results of topic modeling and reviewing the original texts of major precedents by topic showed that the issues of match fixing appearing in the precedents were ‘match fixing route’, ‘Match fixing in relation to E-sports.’, ‘traditional match fixing’, ‘means of identifying match fixing’, and ‘grounds for determining sentencing’. These results can be used as basic data for the development and education of various policies related to match fixing in the future.

토픽모델링을 적용한 코로나19 확산에 따른 스포츠영역 이슈분석

윤효준(Hyo-Jun Yun),박재현(Jae-Hyeon Park),윤지운(JiWun Yoon),전민수(Min-Soo Jeon) 한국체육측정평가학회 2021 한국체육측정평가학회지 Vol.23 No.1

이 연구는 토픽모델링을 활용하여 코로나 확산에 따른 스포츠 영역에 이슈를 분석하기 위한 목적으로 수행되었다. 이를 위해 한국언론진흥재단 언론DB인 빅카이즈를 활용하여, 2020년 1월 6일부터 2020년 12월 31일까지의 스포츠와 코로나의 키워드로 검색되는 국내언론기사를 수집하였다. 수집된 자료 중 중복기사와 관련 없는 기사를 제외하여 최종적으로 2485건의 국내 언론기사를 연구자료로 선정하였다. 분석을 위해 토근화와 명사단어추출, 불필요단어 제거 등의 텍스트 전처리 작업을 하였으며, 토픽수 결정을 위해 토픽수별 Coherence Score를 산출하였다. 산출된 결과를 토대로 토픽수를 6개롤 선정하였다. 토픽모델링은 LDA 알고리즘을 적용하였으며 모든 자료처리는 python 3를 사용하였다. 토픽모델링 결과와 토픽별 주요 언론기사 원문을 검토한 결과, ‘프로스포츠 리그 중단 및 무관중 개막’, ‘정부·지자체·학교 온라인 콘텐츠 활성화’, ‘도쿄올림픽 개최 연기’, ‘정부의 스포츠시설 행정조치’, ‘야외 및 온라인 스포츠 수요 증가’, ‘아마추어 대회개최 연기 및 취소’가 코로나 확산에 따른 스포츠 영역의 이슈임을 확인할 수 있었다. 이러한 결과는 뉴노멀 시대 스포츠 영역에서의 전반적인 논의 주제와 동향을 여론을 통해 파악한다는 점에서 의미가 있을 것이며, 스포츠 관계자 및 정책행위자 현장 문제를 진단하고 개선에 있어 고려해야 할 시사점을 제공할 수 있을 것이다. This study is performed to analyze issues in sports field due to spread of COVID-19 by using topic modeling. To this end, domestic news articles were collected which were searched with key word of sports and COVID-19 from Jan. 6, 2020 to Dec. 31, 2020 by using Big Kinds, DB of the press of Korea Press Foundation. Among collected data, unrelated and overlapped news articles were exempted and finally 2,485 domestic news articles were selected as research material. For analysis, text pre-processing such as tokenization, noun word extraction and unnecessary word removal was performed, and the Coherence Score for each topic number was calculated to determine the number of topics. 6 kinds of topics were selected based on calculated result. LDA algorithm is applied for topic modeling, and python 3 is used for all the data processing. As the result of reviewing the topic modeling result and the original of main news articles for each topic, ‘Discontinuance of professional sports league and Opening without spectators’, ‘Activation of online contents at government·local government·school’, ‘Postponement of holing the Tokyo Olympics’, ’Administrative measures for sports facilities by the government’, ‘Increasing demand for outdoor and online sports’, ‘Postponement and cancellation of holding amateur contest’ are confirmed to be issues due to spread of COVID-19 in sports field. This result has meaning in the sense of grasping general discussion subject and trend in the sports field in the new normal era through the press, and can provide implication to be considered for sports officials and policy actors to diagnose and improve on-site problems.

Monte-Carlo 시뮬레이션을 적용한 이상 배팅 징후 탐지: 남자프로농구승률과 배당률의 관계

이지용(Ji-Yong Lee),박재현(Jae-Hyeon Park),윤지운(JiWun Yoon),윤효준(Hyo-Jun Yun) 한국체육측정평가학회 2020 한국체육측정평가학회지 Vol.22 No.4

이 연구는 남자 프로농구팀 시즌별 승률과 배당률의 관계를 확인하여 이상 배팅 징후 탐지를 통계적으로 접근하고자 하는 목적으로 설계되었다. 실제 경기 자료는 네이버 포털 사이트에서 제공하는 경기 기록을 경기별로 구분하여 자료수집 하였고, 배당률의 경우 국민체육진흥공단에서 위탁하여 운영하고 있는 betman 사이트에서 자료수집을 하였다. 수집된 자료는 MS-Excel 기반 Oracle Crystal Ball ver 1.0 프로그램을 이용하여 monte-carlo 시뮬레이션을 수행하였다. 이 연구에서는 모의자료를 활용해 추정한 승률과 배당률의 관계를 확인하기 위하여 스피어먼의 상관계수 검증을 수행하였다. 첫째, 실제 경기 자료의 평균 안정성을 도모하기 위해 monte-carlo 시뮬레이션을 활용하여 100,000번의 시뮬레이션을 실시하였다. 그 결과 총 8시즌 모두 실제 경기 순위와 모의자료의 순위 간 통계적으로 유의한 정적 상관을 나타냈다. 둘째, 8시즌의 중 총 10번의 사례들이 부적 상관을 나타내지 않는 것으로 나타났다. 이 연구의 결과는 이상 배팅 징후 탐지를 통계적으로 접근하였다는 점에서 의미는 적지 않을 것이며, 승부조작과 관련된 연구 및 교육을 위한 기초차료로서의 의미가 있다고 판단된다. The study was designed to identify the relationship between the winning rate and the dividend rate of the men"s professional basketball team and to detect on unexpected betting. The actual game data was collected by dividing the game records provided by the Naver portal site by game, and the dividend rate was collected from the Betman site, which was entrusted to the Korea Sports Promotion Foundation. By using the Oracle Crystal Ball ver 1.0 program based on MS-Excel, the monte-carlo simulation was executed on the collected data. This study verified Spearman correlation coefficient to identify the relationship between the estimated winning rate and the Sports odds using simulated data. First, 100,000 simulations were conducted using monte-carlo simulations to achieve the average stability of actual match data. As a result, a total of eight seasons showed a statistically significant static correlation between the actual ranking of the competition and the ranking of the simulated data. The results of this study will be significant in that it has detect on unexpected betting, and it is judged to be meaningful as a basis for research and education related to match-fixing.