서지주요정보
Box-jenkins 技法과 다른 短期 時系列 豫測技法과의 실증적 比較 = An empirical comparison of Box-jenkins methods with other short term time series forecasting methods
서명 / 저자 Box-jenkins 技法과 다른 短期 時系列 豫測技法과의 실증적 比較 = An empirical comparison of Box-jenkins methods with other short term time series forecasting methods / 鄭圭錫.
저자명 정규석 ; Chung, Kyu-Suk
발행사항 [서울 : 한국과학기술원, 1980].
Online Access 원문보기 원문인쇄

소장정보

등록번호

4000806

소장위치/청구기호

학술문화관(문화관) 보존서고

MIE 8023

SMS전송

도서상태

이용가능

대출가능

반납예정일

초록정보

This study deals with the three types of Box-Jenkins forecasting method, Box-Jenkins univariate method, Box-Jenkins bivariate method, and Box-Jenkins univariate method with intervention of exogenous variables, and compares the Box-Jenkins univariate method, which is popular and universal, with other two time series forecasting methods, Winter's three parameters linear and seasonal exponential smoothing method and Fourier series method for short term (1-3 months) forecasting. Because it is difficult to use the Box-Jenkins methods, it needs much knowledge and experience to use it effectively following the situation. Box-Jenkins univariate method is universal and popular because of it's simplicity and wider applicability among three methods. The bivariate method can be a good one if there is a nice leading indicator which is closely related with the original series to forecast. The univariate method with intervention is also effective at the time when the intervention effect exists. Here these two methods is compared with univariate method for each cases. Though we deal with the same method to the forecasting of one product, we may obtain different forms of models. And some considerations in dealing with Box-Jenkins methods are introduced, which are obtained through practical application. The subjects above mentioned will help the effective selection and use of Box-Jenkins methods. We must know the difference among some methods to decide which method we will select for real application. And so the comparison of Box-Jenkins univariate method, which is applicable to most products, with Winter's model and Fourier series model is performed in view of forecasting accuracy for 14 products which are chosen at random. Box-Jenkins bivariate method and intervention method are superior to the univariate method in fitting the past data, but they don't have any superiority in forecasting ability as for this study which is a limited one. But they will be powerful for some situations which are suitable for their application. Box-Jenkins methods is superior to Winter's method and Fourier method in forecasting ability by over 10%. And it is especially excellent in case of one month ahead forecasting.

서지기타정보

서지기타정보
청구기호 {MIE 8023
형태사항 [iv], 100 p. : 삽도 ; 26 cm
언어 한국어
일반주기 저자명의 영문표기 : Kyu-Suk Chung
지도교수의 한글표기 : 방석현
학위논문 학위논문(석사) - 한국과학기술원 : 산업공학과,
서지주기 참고문헌 : p. 94-100
주제 Box-Jenkins forecasting.
Forecasting.
시계열. --과학기술용어시소러스
예측 기법. --과학기술용어시소러스
단기 예측. --과학기술용어시소러스
Time-series analysis.
QR CODE qr code