데이터마이닝 기법을 이용한 비만과 과체중의 고위험군 예측모형 구축 : 2005년 국민건강영양조사 중 19세 이상 자료를 중심으로 = Modeling to predict high risk groups of obesity and overweight using data mining technique : based on the over 19 years old data from the third korea national health and nutrition examination survey

다국어 초록 (Multilingual Abstract)

Objectives: The purposes of this study are to examine the prevalence of obesity and overweight in Korea and to suggest a model for high-risk group of obesity and overweight based on data from KNHNESⅢ(Korea National Health and Nutrition Examination Survey) using data mining technique.

Methods: Survey of 4,804 adult men and women(men;1,999, women; 2,805) conducted in 2005 as part of the KNHNES Ⅲ, a nationally representative sample of the Korea population. Body Mass Index (BMI=wt/ht2) was used to define obesity(BMI≥25) and overweight (BMI≥23). Chi-square test and Hochberg's multiple comparison were executed to observe different characteristics among obesity, overweight and normal group with SUDAAN v9.01 Statistics Package Program. Decision tree analysis was used to predict high risk group for obesity and overweight using SAS 9.13. Target variable was BMI and estimate variables were area of residence, sex, age, education, marital status, occupation, type of health insurance, drinking frequency, amount of stress, sleeping time and eating-out frequency.

Results: The prevalence of overweight and obesity was 24.8% 30.9%, respectively. Sex, age, education level, marital status and occupation were different between normal and overweight group. The variables differed between normal and obesity group were area of residence, sex, age, education level, marital status, occupation and sleeping time. The variable differed between overweight and obesity group was sleeping time. The high-risk group of obesity was estimated as men, over 40 years old with white-collar job, based on decision tree analyses. The high risk group of overweight was estimated according to 40-49, 50-59, ≥ 60 years old.

Conclusion: The high risk groups of obesity have been changed according to the socioeconomic development. To overcome Korean adult obesity, the more obesity study is needed using multi-discipline approach. Data mining can be used for finding the high-risk groups of obesity and overweight because data mining is very efficient tool for finding target group. Therefore data-mining can be a useful tool for obesity policy.


Key words: Body Mass Index, Data mining technique, Obesity, Overweight, KNHNESⅢ

목차 (Table of Contents)

• Ⅰ. 서 론 1
• 1. 연구배경 및 필요성 1
• 2. 연구목적 7
• Ⅱ. 연구대상 및 방법 8
• 1. 연구자료 8
• 2. 연구방법 및 분석방법 9
• 3. 연구변수 11
• Ⅲ. 연구결과 13
• 1. 대상자의 일반적인 특성 및 동반질환 13
• 2. 과체중과 비만의 유병률 16
• 3. 일반적 특성에 따른 정상군, 과체중군, 비만군 17
• 4. 데이터마이닝 기법 분석결과 21
• Ⅳ. 고 찰 36
• Ⅴ. 결 론 46
• 참 고 문 헌 48