Earticle

보증데이터의 분석방법은 크게 두 가지로 분류할 수 있다. 첫째, 현상파악이다. 현상 파악은 각 부품별 보증클레임데이터를 이용하여 각부품의 현 수준을 분석하는 방법으 로 단변량 분석방법이다. 즉 부품의 현 수명(신뢰도)를 분석할 수 있는 생명표법, 카랜 마이어 방법이 대표적이라고 할 수 있다. 둘째, 고장원인분석이다. 부품의 고장에 다양 한 인자가 영향을 줄 것이다. 생산부터 고객의 사용조건까지 다양할 것이다. 이처럼 단순히 사용시간을 가지고 분석하는 것이 아닌 다양한 원인변수를 통해서 원인을 파 악하는 다변량 방법이다. 본 연구에서는 다변량 방법 중 Tree기법과 Cox모형을 적용 하고자 한다. 두 방법을 제시하는 이유는 이 두 방법의 서로의 장단점을 보완하여 최 적을 결과를 얻기 위함이다. Tree의 장점은 결과 해석이 다른 모형에 비해 쉽다. 단점 은 유의한 변수가 무엇인지는 알 수 있으나 정량적으로 표현하기 어렵다. 반면에 COX 모형의 경우는 위험도를 정량적으로 표현할 수 있다. 즉 다양한 인자의 기여도를 비교 적 쉽게 찾을 수 있다. 보증데이터는 다양한 정보와 고장에 대한 정보를 가지고 있으 며 이를 바탕으로 원인분석이 가능하다. 본 논문에서는 다양한 인자를 고려하여 고장원인을 추정할 수 있는 다변량 분석방 법을 보증데이터에 적용해 보았고, 실제 유의함을 확인하였다. 특히 Tree 모형 및 Cox 모형를 통해서 서로의 장단점을 보완하였고, 더욱 정확한 원인을 찾을 수 있었다.

The purpose of the warranty data analysis can be classified into two categories. Two goals is a failure cause analysis and life prediction analysis. In this paper first, we applied multivariate analysis method that can be estimated in consideration of various factors on the failure cause warranty data. In particular, we apply the Tree model and Cox model. The advantage of the Tree is easy to interpret this result as compared to other models. In addition Cox model can quantitatively express the risk. Second, this paper proposed a multivariate life prediction model (AFT) considering a variety of factors. By applying the actual warranty data confirmed the usability.

본 연구에서는 국내 하천에서 실측한 유사량 자료를 기초로 데이터 마이닝의 Model Tree 기법을 통해 유사량 산정 공식들을 도출하였으며, 이를 활용하여 내성천에서의 안정하도 단면을 평가하였다. 본 연구에서 도출한 유사량 공식은 국내 모래하상 전체를 대상으로 한 경우, 하폭, 유속, 수심, 경사, 하상토 중앙입경을 선택하였을때 적합도가 가장 높은 것으로 나타났다. 또한 내성천으로 자료의 범위를 한정한 경우에는 경사를 제외한 하폭, 유속, 수심, 하상토 중앙입경을 선택한 유사량 공식의 적합도가 가장 높게 나타났다. 각각의 Model Tree 공식들은 내성천 영주댐 하류 용혈지점에서의 안정하도 단면 평가를 수행하는데 적용되었으며, 현재 내성천의 단면과 비교했을 때 향후 안정하도 단면으로의 변화를 위해 하상의 침식이 발생할 것으로 예측되었다. 또한 현재 하상보다 완만한 경사가 유지될 경우 장기적으로 평형상태에 도달할 것으로 예측되었다.

A Model Tree technique of data mining was applied to derive optimal equations for sediment discharge assessment based on the measured sediment data and then to evaluate stable channel design for Naesung Stream. The sediment discharge formula as a function of channel width, velocity, depth, slope and median grain diameter which was developed by a Model Tree technique with sediment discharge data measured in Korean River had a high goodness-of-fit between measured and calculated results. In case of the sediment discharge formula as a function of channel width, velocity, depth and median grain diameter which was developed by a Model Tree technique with sediment discharge data only measured in Naesung Stream represented the highest goodness-of-fit. Two types of sediment discharge formulas were applied to evaluate stable channel analysis for Yonghyeol Station of Naesung Stream. As a result, bed erosion was expected in the study section compared to the current section. It was also presented that the be slope might be changed to be a milder slope than the current slope to reach equilibrium condition in the long term.

컴퓨터 기반 적응적 검사는 이전에 제시된 문항에 대한 반응을 기반으로 피검사자의 잠재 특질 수준을 추정하기에 가장 적절한 다음 문항을 선택해 출제함으로써 피검사자별 맞춤형 검사를 제시하는 컴퓨터 기반 검사형태다. 컴퓨터 기반 적응적 검사 제작의 핵심 요소 중 하나는 문항 선정 알고리즘이라 할 수 있으며, 최근 결정-트리를 이용한 적응적 심리 검사 구성에 대한 관심과 적용 사례가 늘어나고 있다. 결정-트리는 기계학습 분야에서 주로 연구되어온 예측 모형 중 하나로서 쉽게 해석 가능한 트리-구조를 가진다는 장점에도 불구하고 과적합 문제에 매우 취약하다는 것이 알려져 있다. 본 연구의 목적은 기계학습 분야에서 결정-트리의 대안으로 제시된 앙상블 모형 중에서 해석 가능한 트리-구조를 지닌 Alternating Model Tree (AMT)가 컴퓨터 기반 적응적 심리 검사 제작에 활용될 수 있는지 탐색하는 데 있다. 이를 위해 먼저 AMT의 작동 방식을 적응적 검사의 특징에 비추어 상술하였고, 검사 점수를 예측하는 AMT 기반의 적응적 검사와 결정-트리 기반의 적응적 검사의 예측 성능을 두 개의 심리 검사를 대상으로 비교하였다. 그 결과, AMT는 적응적 검사의 특징을 가지는 것으로 확인되었고, AMT 기반 적응적 심리 검사의 성능은 결정-트리 기반 적응적 심리 검사의 성능과 유사하거나 더 나은 결과를 보였다. 이러한 결과를 바탕으로 본 연구의 의의와 한계, 후속연구에 대한 제언 등을 논의하였다.

Computerized adaptive testing (CAT) is a computer-administered test where the next question for estimating the examinee’s trait level is selected depending on his or her reponses to the previous items, resulting in tailored testing for each individual examinee. A defining feature of CAT stems from its item selection algorithms, among which both research interest and practical applications of decision-tree based CAT (DT-based CAT) have been rising recently. In the field of machine learning, however, it is well known that decision-trees, as a form of predictive models with simple and interpretable tree structures, can be vulnerable to the problem of overfitting or the problem of creating overly complex trees that do not generalize to newly observed data. Among various ensemble techniques developed to adequately address this problem, we the authors paid attention to the Alternating Model Tree (AMT) due to its interpretable tree-like structure. The purpose of this article is to investigate the viability of the Alternating Model Tree (AMT) as an item selection algorithm for constructing CAT. To this end, we first presented a detailed exposition of how AMT-based CAT can be constructed and then compared its performance with DT-based CAT using two sets of publicly available psychological test scores. The results provided supportive evidence that AMT-based CAT is viable, and that AMT-based CAT can predict test scores at least as accurate as DT-based CAT does. Based on our findings, we discuss implications, limitations, and directions of future studies.

Purpose. This study was performed to develop and test a decision-tree model of treatment-seeking behaviors about when Korean patients visit a doctor after experiencing stroke symptoms. Methods. The study used methodological triangulation. The model was developed based on qualitative data collected from in-depth interviews with 18 stroke patients. The model was tested using quantitative data collected from interviews and a structured questionnaire involving 150 stroke patients. The predictability of the decision-tree model was quantified as the proportion of participants who followed the pathway predicted by the model. Results. Decision outcomes of the model were categorized into immediate and delayed treatment-seeking behavior. The model was influenced by lowered consciousness, social-group influences, perceived seriousness of symptoms, past history of hypertension or stroke, and barriers to hospital visits. The predictability of the model was found to be 90.7%. Conclusions. The results from this study can help healthcare personnel understand the education needs of stroke patients regarding treatment-seeking behaviors, and hence aid in the development of educational strategies for stroke patients.

본 연구에서는 유역의 공간상관성을 고려한 다지점 일단위 강수량을 동시에 모의할 수 있는 일강수량 모의기법을 개발하였다. 기존 Hidden Markov Chain Model(HMM)은 단일지점 강수모의에 적용되어 왔으나 관측지점간의 유역상관성을 충분히 고려하지 못하는 문제점을 가지고 있다. 따라서 본 연구에서는 Chow-Liu Tree (CLT) 모형을 적용하여 다변량(multivariate) 형태로써 유역내에 위치한 강우관측소간의 상호종속성을 고려하기 위하여 기존의 동질성 HMM 강우모의기법과 CLT 알고리즘을 결합한 동질성 CLT-HMM 모형을 개발하였다. 본 연구에서 개발된 동질성 CLT-HMM 모형을 사용하여장기간의수문자료를보유하고있는기상청산하의한강유역강수네트워크에대해서 적합성을 검토하였다. 동질성 CLT-HMM 모형을 적용하여 모의된 결과를 보면 일강수량의 계절적 특성뿐만 아니라 일강수량모의 시 강수시계열의 통계적인 특성들까지 우수하게 모의하였다. 추가적으로 상관행렬(correlation matrix)을 이용하여 기상관측소간의 공간상관 재현성을 검토한 결과 관측지점들 사이의 공간상관성도 비교적 우수하게 재현하는 것을 확인할 수 있었다.

This study aims to develop a multivariate daily rainfall simulation model considering spatial coherence across watershed. The existing Hidden Markov Model (HMM) has been mainly applied to single site case so that the spatial coherences are not properly addressed. In this regard, HMM coupled with Chow-Liu Tree (CLT) that is designed to consider inter-dependences across rainfall networks was proposed. The proposed approach is applied to Han-River watershed where long-term and reliable hydrologic data is available, and a rigorous validation is finally conducted to verify the model's capability. It was found that the proposed model showed better performance in terms of reproducing daily rainfall statistics as well as seasonal rainfall statistics. Also, correlation matrix across stations for observation and simulation was compared and examined. It was confirmed that the spatial coherence was well reproduced via CLT-HMM model.

Purpose Climate change affects human health and calls for health meteorological services. The purpose of this study is to find the significant predictors for the demands of the health meteorological information. Methods This study used a descriptive design through structured self-report questionnaires. Data from 956 participants who were at least 18 years old and living in Busan, Korea, were collected from June 1 to July 31, 2009. The data was analyzed using a decision tree method, one of the data mining techniques by SAS 9.1 and Enterprise Miner 4.3 program. Results Two hundred and ninety participants (30.3%) demanded the information, and 505 of them (52.8%) perceived the necessity of health meteorological information. From the decision tree method, the predictors related to the demands of the health meteorological information were determined as “the perception of the necessity of health meteorological information,” “the coping to the weather warnings” and “the importance of the weather forecasting in daily life.” In Particular, the significant different variables in the perception of the necessity of health meteorological information were “female,” “aged over 40” and “environmental diseases.” Thus, the model derived in this study is considered for explaining and predicting the demands of health meteorological information. Conclusions It can be effectively used as a reference model for future studies and is a suggested direction in health meteorological information service and policy development. We suggest health forecasting as a nursing service and a primary health care network for healthier and more comfortable life.

Background: Scapular winging (SW) could be caused by tightness or weakness of the periscapular muscles. Although data mining techniques are useful in classifying or predicting risk of musculoskeletal disorder, predictive models for risk of musculoskeletal disorder using the results of clinical test or quantitative data are scarce. Objects: This study aimed to (1) investigate the difference between young women with and without SW, (2) establish a predictive model for presence of SW, and (3) determine the cutoff value of each variable for predicting the risk of SW using the decision tree method. Methods: Fifty young female subjects participated in this study. To classify the presence of SW as the outcome variable, scapular protractor strength, elbow flexor strength, shoulder internal rotation, and whether the scapula is in the dominant or nondominant side were determined. Results: The classification tree selected scapular protractor strength, shoulder internal rotation range of motion, and whether the scapula is in the dominant or nondominant side as predictor variables. The classification tree model correctly classified 78.79% (p = 0.02) of the training data set. The accuracy obtained by the classification tree on the test data set was 82.35% (p = 0.04). Conclusion: The classification tree showed acceptable accuracy (82.35%) and high specificity (95.65%) but low sensitivity (54.55%). Based on the predictive model in this study, we suggested that 20% of body weight in scapular protractor strength is a meaningful cutoff value for presence of SW.

Purpose: Body image distortion (BID) in adolescents is a crucial problem that causes both abnormal eating habits and unhealthy weight control behaviors. COVID-19 has had a negative impact on adolescents’ psychological and behavioral status, and this might influence the onset of BID in adolescents. This study aimed to develop a prediction model for BID in Korean adolescents using decision tree analysis. Methods: The decision tree analysis was used to develop a prediction model for BID in Korean adolescents using the data of 2021 Korea Youth Risk Behavior Survey Web-based (KYRBS). Results: In the present study, about one-third of the study subjects (31%, n=6,316) showed BID. The BID rate was higher in females (37.2%, node1) than in males (21.2%, node2). Female students with severe GAD-7 level and smartphone use on the weekend over 12h showed the highest rate of BID (66.9%). As to males, the BID rate was the highest (33.1%) among middle school male students who did strength training once a week or none. Conclusion: In order to reduce BID, there is a need to develop a customized BID education and management programs.

Genotype Nisqually-1 is the first model woody plant with an available well-annotated genome. Nevertheless, a simple and rapid transformation of Nisqually-1 remains to be established. Here, we developed a novel shoot regeneration method for Nisqually-1 using leaf petiole and stem segment explants. Numerous shoots formed in the incision of explants within two weeks. The optimized shoot regeneration medium (SRM) contained 0.03 mg l-1 6-benzylaminopurine, 0.02 mg l-1 indole-3-butyric acid and 0.0008 mg l-1 thidiazuron. Based on this, Agrobacterium-mediated genetic transformation of stem explants was examined using the vector pBI121 that contains the β-glucuronidase (GUS) as a reporter gene. Consequently, factors affecting transformation frequency of GUS-positive shoots were optimized as follows: Agrobacteria cell suspension with an OD600 of 0.4, 20 min infection time, 2 days of co-cultivation duration and the addition of 80 μM acetosyringone into Agrobacteria infective suspension and co-cultivation SRM. Using this optimized method, transgenic plantlets of Nisqually-1 – with an average transformation frequency of 26.7% – were obtained with 2 months. Southern blot and GUS activity staining confirmed the integration of the foreign GUS gene into Nisqually-1. This novel transformation system for Nisqually-1 was rapid, efficient, and simple to operate and will improve more genetic applications in this model tree.

This study was carried out to develop individual tree growth model of major coniferous species, Pinus densiflora, Pinus koraiensis, and Larix kaempferi, in Korea. Monitoring data for three year was used in this study and the total number of plots was 48 sites for P. densiflora, 135 sites for P. koraiensis, and 135 sites for L. kaempferi. Individual tree growth of DBH, basal area, and height were analyzed using the correlation coefficients and regressions, and each growth models were developed. The eight kinds of measured tree factors from 1st and 2nd inventory at 3 year intervals were used as independent variables for individual tree growth models: DBH, number of competitors, basal area, site index, stand density index, age, relative height, number of trees per ha. In addition, the competition index calculated through the best method was used as an independent variable. According to the dependent variables, the selected independent variables for the best models were different by species and models. The best models by dependent variables were developed in P. koraiensis. Using the individual tree growth models provided in this study, it is considered to be able to develop the more suitable growth models for P. densiflora, P. koraiensis, and L. kaempferi in Korea.

본 연구의 목적은 2024년 지역사회건강조사의 1인 가구 44,149명의 데이터를 활용하여 식품 불안정성을 예측하는 요인을 식별하는 것이다. 분할 표본 검증을 거친 의사결정나무분석 결과, 1인 가구의 식품 안정성은 경제활동 여부, 교육 수준 및 건강 상태의 복합적인 상호작용과 연관된 것으로 나타났다. 전체 모델에서 식품 안정성 비율이 가장 높은 집단(81.3%)은 높은 주관적 건강 수준, 취업상태, 고등 교육 이수자였다. 반면, 식품 미보장 위험이 가장 높은 집단(8.6%)은 낮은 건강 수준, 낮은 행복감, 미취업 상태와 관련이 있었다. 연령을 뿌리 노드로 설정했을 때, 건강 상태가 나쁜 65세 미만 미취업자에서 식품 미보장률(11.4%)이 가장 높게 나타났으며, 성별을 뿌리 노드로 설정한 경우 행복감이 낮은 미취업 남성(9.1%)이 가장 취약한 것으로 확인되었다. 이러한 결과는 1인 가구의 식품 안정성이 다차원적인 문제임을 시사한다. 위기관리 측면에서, 급증하는 1인 가구의 식품 불안정 위기를 효과적으로 완화하기 위해서는 맞춤형 지원체계를 구축하고 사회적 안정망을 강화하는 것이 필수적이다.

The purpose of this study was to identify predictive factors of food insecurity among single-person households using secondary data from 44,149 participants in the 2024 Korea Community Health Survey. Decision tree analysis, validated through split-sample tests, revealed that food insecurity is determined by the convergence of economic activity, education, and health status. In the unconstrained model, the highest rate of food security (81.3%) was linked to high subjective health, employment, and higher education. Conversely, the highest risk of food insecurity (8.6%) was associated with poor health, low happiness, and unemployment. When age was the root node, unemployed individuals under 65 with poor health showed the highest vulnerability (11.4%). With gender as the root node, unemployed men with low happiness levels exhibited the highest rate of food insufficiency (9.1%). These findings demonstrate that food security in single-person households is a multidimensional issue. It is essential to implement customized support systems and enhance the social safety net to effectively mitigate food insecurity risks for this population.

본 연구는 질병관리청에서 제공하는 2021년 청소년건강행태조사 데이터를 활용하여 의사결정나무 분석 법으로 코로나19 시대 중 ‧ 고등학생의 성별에 따른 체질량지수를 예측하는 모델을 구현하고자 하였다. 예측 모형에서 체질량지수는 성별에 따라 유의미한 차이가 있었고, 남자 중학생의 경우에 고강도 신체활 동을 전혀 하지 않으면서 주말 동안의 스마트폰 사용 시간이 11시간 이상인 경우 비만율이 37.4%로 가장 높게 나타났다. 남자 고등학생은 고강도 신체활동을 하지 않거나 일주일에 1번 정도 하면서 학업 성적이 중간-낮음의 수준인 경우 비만 비율이 36.2%로 가장 높았다. 여중생의 경우 학업성적이 낮으면서 불안장애수준이 심각하거나 반대로 완전히 낮은 경우 비만 비율이 18.4%로 가장 높았다. 여고생의 경우 는 코로나19에 따른 경제적 상황에 크게 변화가 있으면서 주말 동안의 스마트폰 사용 시간이 9.8시간 이상인 경우에 비만율이 20.5%로 가장 높았다. 코로나19 상황 이후 비만 청소년 관리를 위하여 개인적 특성 및 상황을 고려한 맞춤형 관리프로그램의 개발과 적용이 시급하다.

This study built a prediction model for BMI by sex of middle and high school students in the COVID-19 era by using 2021 Korea Youth Risk Behavior Survey. This study used data from 30,015 middle school and 24,833 high school students. The data was analyzed by using decision tree analysis. For male middle school students, if they did not engage in strenuous physical activity and spent more than 11 hours on a smartphone during the weekend, the obesity rate was the highest at 37.4%. Male high school students, those who did 0-1time strenuous physical activity a week and had medium-low level academic performance showed the highest level of obesity(36.2%). Middle school girls, when the academic performance was low and the anxiety disorder level was severe or conversely, the obesity rate was the highest at 18.4%. High school girls, the obesity rate was the highest at 20.5% when there was a significant change in the economic situation due to COVID-19 and the time spent on smartphone over the weekend was 9.8 hours or more. It is needed to develop a customized management and counseling program for obese adolescents.

Purpose: The aim of this study was to predict the subgroups vulnerable to poorer health-related quality of life (HRQoL) according to gender in older adults. Methods: Data from 5,553 Koreans aged 65 or older were extracted from the Korea National Health and Nutrition Examination Survey. HRQoL was assessed using the EQ-5D tool. Complex sample analysis and decision-tree analysis were conducted using SPSS for Windows version 27.0. Results: The mean scores of the EQ-5D index were 0.93 ± 0.00 in men and 0.88 ± 0.00 in women. In men, poorer HRQoL groups were identified with seven different pathways, which were categorized based on participants’ characteristics, such as restriction of activity, perceived health status, muscle exercise, age, relative hand grip strength, suicidal ideation, the number of chronic diseases, body mass index, and income status. Restriction of activity was the most significant predictor of poorer HRQoL in elderly men. In women, the poorer HRQoL groups were identified with nine different pathways, which were categorized based on participants’ characteristics, such as perceived health status, restriction of activity, age, education, unmet medical service needs, anemia, body mass index, relative hand grip, and aerobic exercise. Perceived health status was the most significant predictor of poorer HRQoL in elderly women. Conclusion: This study presents a predictive model of HRQoL in older adults according to gender and can be used to detect individuals at risk of poorer HRQoL.

본 연구에서는 데이터 마이닝을 이용하여 인구통계학적 변인, 신체․생리적 변인, 흡연행동 변인, 및 심리적 변인들로 2년 동안의 금연 유지 요인을 탐색하려고 하였다. 본 연구의 대상자는 입원 금연 프로그램에 참여한 73명이었는데, 이들의 평균 연령은 44.89(SD=9.61)세 이었으며, 남성이 65명이고 여성이 8명이었다. Fagerstrom의 니코틴 의존도 검사, 아이젱크 성격검사, 다차원적 대처 척도, 및 Jerusalem과 Schwarzer의 자기효능감 척도로 얻은 정보와 다양한 의학적 검사의 측정치가 분석에 포함되었다. 43가지 변인들을 동시에 투입하여 금연 지속기간을 예측하는 의사결정 나무분석을 실시하였다. 분석 결과, 적극적 대처, 신경증적 경향성, 정신병적 경향성자기비판, 사회적 지원 추구, 고밀도 콜레스테롤(HLD), 및 혈당 수준이 2년 동안의 장기적 금연 유지를 예측하는 가장 영향력 있는 요인으로 나타났다. 2년 동안 추적한 결과, 적극적으로 대처하고 신경증적 경향성과 정신병적 경향성이 낮은 사람들은 평균 646.25일 동안 금연을 유지한 반면에, 적극적으로 대처하지 않고 고밀도 콜레스테롤 수치가 낮으면서 사회적 지원을 추구하지 않는 사람들은 평균 17.83일 동안만 금연을 유지했다. 본 연구를 통해 금연의 장기적 유지에 생물․심리․사회적 변인이 어떤 역할을 하는지 확인할 수 있었다.

This study aims to investigate predictors of long-term (2 years) abstention from smoking with demographic profiles, physical and physiological variables, smoking behaviors, and psychological variables. Participants were 73 smokers (65 males, 8 females) participated in a hospitalized smoking cessation program whose mean age was 44.89 (SD=9.61). Participants completed questionnaires and psychological tests including: Fagerstrom Test for Nicotine Dependence, Eysenck Personality Inventory, Multidimensional Coping Scale, and Jerusalem and Schwarzer's Self-Efficacy Scale. And data from variety of clinical measurements were included for analysis. To identify participants abstention, researchers followed them for 2 years. Data Mining Decision Tree was performed with 43 variables by Answer Tree 3.0v. Decision Tree model included significant variables for abstention as predictors, such as active coping, neuroticism, psychoticism, self-criticism, High Density Lipoprotein (HLD), seeking social support, level of glucose. Decision Tree model showed that those who coped actively, were less neurotic and psychotic, had abstained for average of 646.25 days/2 years, while those who coped unactively, had low level of HDL, were unlikely to sought social support, had abstained only for average of 17.83 days/2 years. These results suggest roles of biopsychosocial variables for long-term successful smoking cessation.

문항반응나무모형(Item Response Tree: IRTree)은 응답자의 설문 응답 시 발생할 수 있는 인지적응답 반응 과정을 나무구조로 반영하여 문항반응이론에 접목시킨 모형으로, 최근 응답 편향의일종인 극단응답반응양식(Extreme Response Style: ERS)과 관련한 연구에서 활발하게 활용되고 있다. 나아가 ERS로 인해 발생할 수 있는 자료의 이질성을 탐색하는 방법으로 IRTree 모형에 잠재집단모형을 추가한 혼합문항반응나무모형(Mixture Item Response Tree: MixIRTree)을 제안되고있다. 본 연구에서는 MixIRTree 모형의 적용 가능성을 탐색하기 위하여 한국청소년정책연구원의 한국 아동·청소년 패널조사의 부적정서 문제 중 신체적 증상과 우울 척도를 활용하였다. 연구 결과에 따르면 서로 다른 수준의 ERS를 보이는 세 개의 잠재집단을 확인할 수 있었다. 첫 번째로, 잠재집단 3은 잠재집단 1과 2에 비하여 적은 ERS 경향을 보였고, 잠재집단 1이 모든 척도에서 가장 강한 ERS 경향을 보였다. 두 번째로, ERS 경향을 보인 잠재집단 1과 2에서는 척도에 따라서 ERS 경향이 다르게 나타났다. 잠재집단 1은 우울 척도에서 상대적으로 더강한 ERS 경향을 보였으며, 잠재집단 2는 신체적 증상 척도에 상대적으로 더 강한 ERS를 보이는 것으로 나타났다. 이와 같은 결과를 바탕으로, 본 연구에서 제안된 MixIRTree와 같은 심리측정모형을 활용하여 잠재목표특성치와 ERS를 동시에 측정하면서 ERS로 발생할 수 있는 잠재집단 간 차이를 탐색하는 것이 필요함을 경험적 자료 분석을 통하여 시사하였다.

Item response tree model (IRTree) is a sort of a multidimensional item response theory that incorporates a tree structure depicting a respondent's cognitive response process into item response theory. IRTree has recently received considerable attention from researchers for measuring extreme response style (ERS) that can threaten reliability and factor structure in self-report survey data. In addition to the work of IRTree, we extend IRTree as the mixture item response tree model (MixIRTree), which combines latent class analysis with IRTree for measuring and investigating the possible heterogeneity of data caused by ERS. For analyses, this study used the body symptom and depression scales of the emotional problems in the third wave of KCYPS 2018. The results of the study are as follows. First, the three latent classes were captured, showing the different levels of ERS. The latent class 3 did not show the ERS tendency than the latent classes 1 and 2. The latent class 1 showed the largest ERS tendency. Second, the ERS tendency across the latent classes was different between the negative affect scales. For example, the ERS tendency of the latent class 1 was larger in the depression scale than body symptom scale, while the ERS tendency of the latent class 2 was larger in the body symptom scale than the depression scale. Based on these findings, we provide practical implications of the MixIRTree for measuring the heterogeneity of ERS and discuss the future direction for further study.

본 논문에서는 싱글 야외 영상에서 날씨 분류를 위한 계층적 이미지 트리 모델을 정의하고, 영상의 밝기와 k-평균 세분화 영상을 이용한 날씨 분류 알고리즘을 제안하였다. 계층적 이미지 트리 모델의 첫 번째 레벨에서 실내와 야외 영상을 구분하고, 두 번째 레벨에서는 야외 영상이 주간, 야간 또는 일출/일몰 영상인지를 밝기 영상과 k-평균 세분화 영상을 이용하여 판단하였다. 마지막 레벨에서는 두 번째 레벨에서 주간 영상으로 분류된 경우 에지 맵과 안개 율을 기반으로 맑은 영상 또는 안개 영상인지를 최종 추정하였다. 실험 결과, 날씨 분류가 설계 규격대로 수행됨을 확인할 수 있었으며, 제안하는 방법이 주어진 영상에서 효과적으로 날씨 특징이 검출됨을 보였다.

In this paper, a hierarchical image tree model for weather classification is defined in a single outdoor image, and a weather classification algorithm using image intensity and k-mean segmentation image is proposed. In the first level of the hierarchical image tree model, the indoor and outdoor images are distinguished. Whether the outdoor image is daytime, night, or sunrise/sunset image is judged using the intensity and the k-means segmentation image at the second level. In the last level, if it is classified as daytime image at the second level, it is finally estimated whether it is sunny or foggy image based on edge map and fog rate. Some experiments are conducted so as to verify the weather classification, and as a result, the proposed method shows that weather features are effectively detected in a given image.

본 연구는 잠재혼합된 청소년의 신체건강행위에 관여하는 가족 내재적⋅외재적 요소에 대한가지경로 분석을 통하여, 가족경계를 넘어 발달영역을 확장해 가는 청소년의 신체건강을 제고하기 위하여 이루어졌다. 혼합분포된 청소년의 신체건강행위에 대한 보다 과학적인 분석을 실시하기 위하여 이에 적합한 가우시안믹스처모델(GMM)을 활용하였다. 그리고 가지경로를 분석하기 위해 Decision Tree모형을 활용하였다. jamovi 및 R(ver.4.1.0)/R-studio 분석도구로 활용하여 한국아동⋅청소년패널조사2018 패널자료의 2차년도(2019년) 데이터 자료(중학교 1학년2,590명과 보호자 2,590)를 분석하였고 연구결과는 다음과 같다. 첫째, GMM을 통하여 청소년들의 신체건강행위에 대해서 잠재혼합된 두 개의 상이한 집단으로 분류되었으며, 건강행위 군집과 불건강행위 군집으로 명명하였다. 둘째, 두 군집에 대한 t 검정 결과, 건강행위 군집이 가족 외재적 요소인 청소년활동 및 교외동아리활동 그리고 가족내재적 요소인 어머니 근로여부를 제외하고 모두 유의하게 높았다. 셋째, 청소년의 신체적 건강에 대해서 건강행위 군집과 불건강행위 군집을 분류하는 가지경로 분석을 하기 위하여 디시즌트리를 실시한 결과, 가족 외재적 요소인 친구관계와 교사관계가 긍정적이며, 가족 내재적요소인 어머니의 학력이 고학력일 경우 건강행위 군집으로 분류예측 되었다.

This study was conducted to improve the physical health of adolescents who are expanding their developmental domain beyond family boundaries through branch path analysis on the factors involved in the physical health behavior of latent mixed adolescents. In order to conduct a more precise analysis of the mixed-distributed physical health status of mixed-distributed adolescents, an appropriate Gaussian Mixture Model(GMM) was used. And a decision tree model was used to analyze branch paths. Using jamovi and R(ver.4.1.0)/R-studio as an analysis tool, the second year (2019) data of the Korean Children and Youth Panel Survey 2018 panel data (2,590 first-year middle school students and 2,590 guardians) were analyzed. The research results are as follows. First, through GMM, the physical health of adolescents was classified into two different groups that were latent mixed, and they were named the healthy behavior cluster and unhealthy behavior cluster. Second, as a result of t-test for both clusters, the healthy behavior cluster was significantly higher except for youth activities and suburban club activities, which are factors outside the family, and whether or not the mother worked, which is an intrinsic factor in the family. Third, as a result of the decision tree to analyze the branch path that classifies healthy behavior and unhealthy behavior clusters of adolescents, the friend relationship and teacher relationship, which are external factors of the family, are positive, and the mother’s educational background, which is an intrinsic factor of the family, is positive. This high educational level was predicted to be classified as a healthy behavior cluster.

안개는 대체수자원이 될 수 있으나 교통사고 위험을 높이고 공항 운영에 제약을 가하는 사회적 영향이 큰 기상현상이다. 본 연구에서는 1 km 미만 가시거리(시정)로 정의되는 안개 발생을 기상자료로 예측하는 지역 기계학습모형을 개발하고 그 예측력을 평가하였다. 전라북도 지역의 10개 기상청 지상관측소의 2017-2019년 시정 및 기상관측자료로 앙상블 분류기법인 Extreme Gradient Boosting (XGB), Light Gradient Boosting(LGB), Random Forests (RF)를 학습시켜 지역 안개 모형을 개발하였고 독립적인 2020년 자료로 모형의 사용성을 평가하였다. 그 결과, 학습·검증기간(2017-2019)에는 True Skill Score를 기준으로 가장 높은 예측력을 보인 방법은 LGB 기법이었지만 다른 두 모형에 비해 False Alarm Ratio가 컸다. RF 모형과 XGB 방법 역시 기존 연구에 상응하는 예측성능을 보이는 것으로 확인되었다. 2020년 자료를 입력해 안개 발생을 모의했을 때 세 모형의 예측성능은 2017-2019년 기간보다 떨어졌지만 모두 관측 안개일수의 공간분포와 일관되는 안개 위험을 예측했다. 세 기계학습모형은 안개위험이 상대적으로 높은 지역을 추출하는 기법으로 사용이 가능할 것으로 보인다.

While it could become an alternative water resource, fog could undermine traffic safety and operational performance of infrastructures. To reduce such adverse impacts, it is necessary to have spatially continuous fog risk information. In this work, tree-based machine-learning models were developed in order to quantify fog risks with routine meteorological observations alone. The Extreme Gradient Boosting (XGB), Light Gradient Boosting (LGB), and Random Forests (RF) were chosen for the regional fog models using operational weather and visibility observations within the Jeollabuk-do province. Results showed that RF seemed to show the most robust performance to categorize between fog and non-fog situations during the training and evaluation period of 2017-2019. While the LGB performed better than in predicting fog occurrences than the others, its false alarm ratio was the highest (0.695) among the three models. The predictability of the three models considerably declined when applying them for an independent period of 2020, potentially due to the distinctively enhanced air quality in the year under the global lockdown. Nonetheless, even in 2020, the three models were all able to produce fog risk information consistent with the spatial variation of observed fog occurrences. This work suggests that the tree-based machine learning models could be used as tools to find locations with relatively high fog risks.