View : 25 Download: 0

Comparison of Prediction Models for Analyzing Smartphone Addition of Korean Adolescents

Title
Comparison of Prediction Models for Analyzing Smartphone Addition of Korean Adolescents
Other Titles
한국 청소년들의 스마트폰 중독 예측을 위한 분석 모형 비교
Authors
김다예
Issue Date
2017
Department/Major
대학원 통계학과
Publisher
이화여자대학교 대학원
Degree
Master
Advisors
이동환
Abstract
According to a survey by the female family department, the domestic penetration rate of smartphones has exceeded 80%. According to the results of the Future Creative Science Division together with the Korea Information Technology Promotion Agency, based on 2015, about 2.4% of smartphone users are high risk and 13.8% belong to a potential risk group. This is an increased figure than last year, and the social concern for smartphone poisoning in the future is expected to further increase. Especially, the poisoning tendency can be fatal as the age is lower, so youth smartphone poisoning can become a social problem. Therefore, in this research, we investigate how mental health condition of young people affects adolescent smartphone poisoning. In addition, we will construct a statistical model that can predict young people's smartphone poisoning and compare whether some models' predictive power is good. The analysis utilized the mental health survey data for 714 students in the junior high school located in Seoul in psychiatry at Seoul Polramé hospital. I examined the mental health scale that affects youth 's smartphone poisoning. In the analysis method for the prediction model, we used the MSE value by using linear regression, GAM, Support Vector Machine, neural network method, and compared which model has the best predictive power.;본 논문에서는 청소년들의 스마트폰 중독에 대한 특징에 대하여 살펴본 후, 여러 개의 예측 모형을 만들어 예측력을 비교했다. 제안된 모형 중 예측력이 가장 좋은 모형을 선택하여 어떤 것이 중요 변수로 사용되었는지 모형별로 살펴보았다. 그래서 청소년의 스마트폰 중독과 관련하여 성별로 차이가 있으며, SVR 모형의 예측력이 좋고 인터넷 중독 성향, 불안성과 공격적인 성향 등 변수가 높은 예측 기여도를 가짐을 알 수 있다.
Fulltext
Show the fulltext
Appears in Collections:
일반대학원 > 통계학과 > Theses_Master
Files in This Item:
There are no files associated with this item.
Export
RIS (EndNote)
XLS (Excel)
XML


qrcode

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

BROWSE