• 응용통계연구
• /
• 제21권4호
• /
• pp.639-647
• /
• 2008

본 연구는 국민건강보험공단의 건강검진데이터, 자격 및 보험료 그리고 진료비 데이터를 활용하여 고혈압 관리를 위한 맞춤형 고혈압 사후관리모형(고혈압 진료예측모형 및 고혈압 진료순응도세분화모형)을 개발하고자 하였다. 모형 개발에는 데이터마이닝의 로지스틱 회귀모형, 의사결정나무 그리고 앙상블 모형을 활용하였다. 고혈압 진료예측모형에서는 3가지 모형 중 로지스틱 회귀모형이 가장 우수한 모형으로 채택되었으며, 고혈압 진료순응도세분화모형은 의사결정나무모형을 통해 개발되었다. 본 연구는 전국 규모의 수년간 축적된 자료를 데이터마이닝을 활용함으로써 고혈압의 진료 및 진료순응도에 이르는 고혈압 사후관리 프로세스 전반에 걸친 결과를 도출함으로써 우리나라 고혈압 사후관리체계 구축에 기여할 것으로 사료된다.

• 한국사회복지학
• /
• 제39권
• /
• pp.5-40
• /
• 1999

이 연구의 목적은 실업자의 실업기간과 재취업 형태에 영향을 미치는 주된 요인이 무엇인지를 파악하고, 그에 기초하여 보다 바람직한 고용관련정책 마련에 필요한 기초자료를 제공하는 데 있다. 이를 위해 이 연구는 보건사회연구원과 노동연구원이 공동으로 실시한 '1998년도 전국 실업실태 및 복지욕구조사' 자료를 Weibull 모델과 Logistic모델을 이용하여 분석하였다. 분석한 결과, 전체 실업자를 대상으로 할 경우, 연령, 가구 월 평균소득, 가구주 여부, 결혼 여부, 이전 직장의 규모, 실직사유 등의 변수가 실업자의 실업기간에 영향을 미치는 주요 요인으로 파악되었고, 연령, 이전 직장의 직종과 근무형태, 고용보험가입 여부 등의 변수가 실업자의 재취업형태에 영향을 미치는 주요 요인으로 파악되었다. 이러한 실증적 분석의 결과를 바탕으로 이 연구는 불안정한고용상태와 실업, 그리고 실업과 불안정한 재취업의 순환을 반복할 수 있는 취약계층 근로자를 위한 보다 체계적인 실업대책의 필요성을 논하면서 정책적인 제언을 하고자 한다.

• Journal of the Korean Data and Information Science Society
• /
• 제23권2호
• /
• pp.235-245
• /
• 2012

로지스틱판별분석은 금융 분야에서 유용하게 사용되고 있는 통계적 기법으로 신용평가 시 해석이 쉽고 우수한 분별력으로 많이 활용되고 있지만 종속변수에 대한 설명변수들의 비선형적인 관계를 설명하는 부분에는 한계점이 있다. 일반화가법모형은 로지스틱판별모형의 장점과 함께 종속변수와 설명변수 사이의 비선형적인 관계도 설명할 수 있다. 그러나 연속형 설명변수의 수가 대단히 많은 경우이 두 방법은 모형에 유의한 변수를 선택해야하는 문제점이 있다. 따라서 본 연구에서는 다수의 연속형 설명변수들을 공통요인분석자혼합모형에 의한 차원축소를 통해 변환된 소수의 요인점수들을 일반화가법모형의 새로운 연속형 설명변수로 사용하여 신용분류를 하는 방법을 제시한다. 실제 금융자료를 이용하여 로지스틱판별모형과 일반화가법모형, 그리고 본 연구에서 제안한 방법에 의한 정분류율을 비교한 결과 본 연구에서 제안한 방법의 분류 성능이 더 우수하였다.

• 한국안전학회지
• /
• 제28권1호
• /
• pp.95-101
• /
• 2013

This study tried to develop a basis for quantitative index of working postures associated with WMSDs(Work-related Musculoskeletal Disorders) that could overcome realistic restriction during application of typical checklists for WMSDs evaluation. The baseline data for this study was obtained from automobile manufacturing company(A total of 603 jobs were observed). Specifically, data for shoulder postures was analyzed to have a better and more objective method in terms of job relevance than typical methods such as OWAS, RULA, and REBA. Major statistical tools were Clustering, Logistic regression and so on. The main results in this study could be summarized as follows; 1) The relationships between working postures and WMSDs symptoms at shoulder were statistically significant based on the results from logistic regression. 2) Based on clustering analysis, three levels for WMSDs risk at shoulder were produced for both flexion and abduction were statistically significant. Specific results were as follows; Shoulder flexion: low risk(< $37.7^{\circ}$), medium risk($37.7^{\circ}{\sim}70.0^{\circ}$), high risk(> $70.0^{\circ}$) Shoulder abduction: low risk(< $26.5^{\circ}$), medium risk($26.5^{\circ}{\sim}56.8^{\circ}$), high risk(> $56.8^{\circ}$). 3) The sensitivities on risk levels of shoulder flexion and abduction were 64.0% and 20.6% respectively while the specificities on risk levels of shoulder flexion and abduction were 99.1% and 99.3% respectively. The results showed that the data associated with shoulder postures in this study could provide a good basis for job evaluation of WMSDs at shoulder. Specifically, this evaluation methodology was different from the methods usually used at WMSDs study since it tried to be based on direct job relevance from real working situation. Further evaluation for other body parts as well as shoulder would provide more stability and reliability in WMSDs evaluation study.

• International Journal of Computer Science & Network Security
• /
• 제22권5호
• /
• pp.294-302
• /
• 2022

Detection of fake news is a complex and a challenging task. Generation of fake news is very hard to stop, only steps to control its circulation may help in minimizing its impacts. Humans tend to believe in misleading false information. Researcher started with social media sites to categorize in terms of real or fake news. False information misleads any individual or an organization that may cause of big failure and any financial loss. Automatic system for detection of false information circulating on social media is an emerging area of research. It is gaining attention of both industry and academia since US presidential elections 2016. Fake news has negative and severe effects on individuals and organizations elongating its hostile effects on the society. Prediction of fake news in timely manner is important. This research focuses on detection of fake news spreaders. In this context, overall, 6 models are developed during this research, trained and tested with dataset of PAN 2020. Four approaches N-gram based; user statistics-based models are trained with different values of hyper parameters. Extensive grid search with cross validation is applied in each machine learning model. In N-gram based models, out of numerous machine learning models this research focused on better results yielding algorithms, assessed by deep reading of state-of-the-art related work in the field. For better accuracy, author aimed at developing models using Random Forest, Logistic Regression, SVM, and XGBoost. All four machine learning algorithms were trained with cross validated grid search hyper parameters. Advantages of this research over previous work is user statistics-based model and then ensemble learning model. Which were designed in a way to help classifying Twitter users as fake news spreader or not with highest reliability. User statistical model used 17 features, on the basis of which it categorized a Twitter user as malicious. New dataset based on predictions of machine learning models was constructed. And then Three techniques of simple mean, logistic regression and random forest in combination with ensemble model is applied. Logistic regression combined in ensemble model gave best training and testing results, achieving an accuracy of 72%.

• 산업진흥연구
• /
• 제7권1호
• /
• pp.1-8
• /
• 2022

본 논문은 중도중단 데이터가 포함된 생존데이터의 경우 적용할 수 있는 기계학습 방법에 대해 살펴보았다. 우선 탐색적인 자료분석으로 각 특성에 대한 분포, 여러 특성들 간의 관계 및 중요도 순위를 파악할 수 있었다. 다음으로 독립변수에 해당하는 여러 특성들과 종속변수에 해당하는 특성(사망여부) 간의 관계를 분류문제로 보고 logistic regression, K nearest neighbor 등의 기계학습 방법들을 적용해본 결과 적은 수의 데이터이지만 통상적인 기계학습 결과에서와 같이 logistic regression보다는 random forest가 성능이 더 좋게 나왔다. 하지만 근래에 성능이 좋다고 하는 artificial neural network나 gradient boost와 같은 기계학습 방법은 성능이 월등히 좋게 나오지 않았는데, 그 이유는 주어진 데이터가 빅데이터가 아니기 때문인 것으로 판명된다. 마지막으로 Kaplan-Meier나 Cox의 비례위험모델과 같은 통상적인 생존분석 방법을 적용하여 어떤 독립변수가 종속변수 (t_i, δ_i)에 결정적인 영향을 미치는지 살펴볼 수 있었으며, 기계학습 방법에 속하는 random forest를 중도중단 데이터가 포함된 생존데이터에도 적용하여 성능을 평가할 수 있었다.

• 보건의료산업학회지
• /
• 제8권3호
• /
• pp.147-155
• /
• 2014

The objective of the study was to analyze the regional variance of late preterm birth (LPT: 34-36 weeks) by analyzing 2008-2012 birth certificated data of seven metropolitan cities (536,984 births: primiparous singleton birth) from Korea Statistics. The odds ratio and 95% confidence intervals were calculated from multinominal logistic regression analyses to describe the regional variance of LPT adjusted for maternal and infantile variables. The highest incidence of LPT rate by region were observed in Ulsan metropolitan city (3.7 percent), and the lowest in Deajon metropolitan city (3.1 percent). After adjustment by logistic regression for infantile sex, maternal variables, there was a significant increase in the risk of late preterm birth in Ulsan metropolitan city (odds ratio: 1.21) as compared with the incidence of LPT in Deajon metropolitan city. The odds ratio of LPT by region were 1.17 in Daegu metropolitan city, 1.13 Busan metropolitan city, and 1.12 in Incheon metropolitan city. More research is required to understand the risk factors for late preterm birth in this area including socio-demographic factors, medical factors, and regional and environmental factors.

• 보건교육건강증진학회지
• /
• 제23권3호
• /
• pp.65-83
• /
• 2006

Objectives: This is one of the first efforts to describe incidence of alcohol-related problems and to identify environmental correlates associated with them among colleges. Methods: Date were collected by a sample of 105 college administrators who are in charge of student affairs in colleges nationwide through self-administrated questionnaire. Both logistic and linear multiple regression analyses were employed to identify the correlates associated with alcohol-related problems. Results: Most of colleges(76.6%) under study reported to have at least one alcohol-related problem in previous years. Interpersonal violence was alcohol-related problem taken placed most frequently, followed by making noise episode, having property damaged and motor vehicle accidents. Logistic regression analysis identified factors associated with incidents of alcohol related problems. They included being private colleges, numbers of prevention activities, product promotion and marketing by alcohol industry and alcohol accessibility to drinking context. Multiple regression analyses showed that correlates associated with numbers of alcohol-related problems included being a private college, being located in rural area, having drinking density, product promotion and availability of alternative activities to drinking. Conclusions: Environmental correlates were associated with incidence of alcohol related problems in colleges nationwide. Policy implications were discussed.

• 한국신뢰성학회지:신뢰성응용연구
• /
• 제17권1호
• /
• pp.38-49
• /
• 2017

Purpose: The purpose of this study is to introduce the assessment model for industrial accidents prevention policy. Methods: 10 years of industrial accidents data are explored through EDA approach. Case control study is tried in order to assess the effectiveness of the measures taken by Korea Safety and Health Agency, Civilian, and 'Ministry of Employment and Labor'. Propensity score matching is used to match the characteristics of the two groups compared, and then case control study is again conducted. Next, logistic and Poisson regressions are used to assess the risk factors. Results: According to case control study involvement of 'Korea Safety and Health Agency' and 'Ministry of Employment and Labor' were not effective, but Civilian was. Propensity score matching leads to the same conclusion. Poisson regression reveals the impact of the risk factors on the industrial accidents. Industrial accidents occur more often as the number of employees grows. Mining, farming, fishing, 'transportation storage and telecommunication' and forestry have a higher level of industrial accidents but service industry has a lower level. It is odd that more involvement of Korea Safety and Health Agency, Civilian, and Ministry of Employment and Labor means more industrial accidents. Conclusion: 'Korea Safety and Health Agency', Civilian, and 'Ministry of Employment and Labor' seem to visit those industries with more industrial accidents.

• 통상정보연구
• /
• 제10권4호
• /
• pp.301-319
• /
• 2008

The development of Korea in the Northeast through trade is not a matter of choice as far as national strategy is concerned, but is an important national policy that is a matter of life or death which will determine the future fate of Korea. This thesis will attempt to arrive at a general, tangible plan for the development of Korea in the Northeast centered around trade by examining the change in naval environment at home and abroad, the given economic situation in the Northeast, and the present state of essential port development in Korea-China-Japan. Its objective will be to provide strategies for the development of Busan port confrontation. For the last ten years, China has been growing at a rapid rate. Since a lot of the volume of naval trade is being transferred from Korea to China, we must do everything we can to improve the service and reduce cost. In addition, Japan also is losing international position. Japan's government and the private industry are trying to make Super Core Ports a prominent feature of their port system. If the Busan port system is to remain competitive, these aspects of the Japanese port system must be kept in mind to prevent trade from going to other ports with more competitive systems.