• Journal of Korea Water Resources Association
• /
• v.39 no.12 s.173
• /
• pp.997-1012
• /
• 2006

We have presented a nonparametric stochastic approach for the SOI(Southern Oscillation Index) series that used nonlinear methodology called Nonlinear AutoRegressive(NAR) based on conditional kernel density function and CAFPE(Corrected Asymptotic Final Prediction Error) lag selection. The fitted linear AR model represents heteroscedasticity, and besides, a BDS(Brock - Dechert - Sheinkman) statistics is rejected. Hence, we applied NAR model to the SOI series. We can identify the lags 1, 2 and 4 are appropriate one, and estimated conditional mean function. There is no autocorrelation of residuals in the Portmanteau Test. However, the null hypothesis of normality and no heteroscedasticity is rejected in the Jarque-Bera Test and ARCH-LM Test, respectively. Moreover, the lag selection for conditional standard deviation function with CAFPE provides lags 3, 8 and 9. As the results of conditional standard deviation analysis, all I.I.D assumptions of the residuals are accepted. Particularly, the BDS statistics is accepted at the 95% and 99% significance level. Finally, we split the SOI set into a sample for estimating themodel and a sample for out-of-sample prediction, that is, we conduct the one-step ahead forecasts for the last 97 values (15%). The NAR model shows a MSEP of 0.5464 that is 7% lower than those of the linear model. Hence, the relevance of the NAR model may be proved in these results, and the nonparametric NAR model is encouraging rather than a linear one to reflect the nonlinearity of SOI series.

• Communications for Statistical Applications and Methods
• /
• v.1 no.1
• /
• pp.131-140
• /
• 1994

Box-Jenkins 시계열 분석에서 모형검진을 위한 통계량으로 잔차의 자기상관함수를 이용한 Box와 Pierce(1970)의 포트맨토우 검정과 Ljung과 Box(1978)의 변형된 포트맨토우 검정을 Basawa(1987)가 제안한 예측오차를 이용한 모형 검진 방법과 비교, 분석하였다. 시뮬레이션 연구를 수행하여 경험적 평균, 분산 및 유의 수준을 비교하여 과대적합의 방법을 이용하여 검정력을 비교하였다.

• Communications for Statistical Applications and Methods
• /
• v.16 no.4
• /
• pp.587-594
• /
• 2009

In this paper the change point problem of the error terms in linear regression models is considered. Since fixed or stochastic independent variables and weakly dependent errors are assumed, usual multiple regression models and time series models including ARMA are covered. We use the estimates of probability density function based on residuals in order to test the distribution change of the unobserved errors. Under some mild conditions, the test using the residuals is proved to have the same limiting distribution as the test based on true errors.

• Journal of the Korean Data and Information Science Society
• /
• v.12 no.1
• /
• pp.27-40
• /
• 2001

The purpose of this study is to test a multiple change point in the regression model with the passage of time, using the estimated residual variance figure suggested by Gasser, Sroka and Jennen - Steinmez (GSJS). As a result of the simulation, it is showed that there is a jump change of the estimated residual variance figure at that time of change point. The way to analyse a intuitive multiple change point through graphics is more effective and accurate than any other existing ways.

• The Korean Journal of Applied Statistics
• /
• v.30 no.3
• /
• pp.427-439
• /
• 2017

Quade (1967) proposed RANK ANCOVA, which is a nonparametric method to test differences between treatments when there are covariates. Hwang and Kim (2012) also proposed a joint placement test on covariate-adjusted residuals. In this paper, we proposed a new nonparametric method to control the effect of covariate on a response variable that uses linear statistics on covariate adjusted-residuals. The score function used in the linear statistics was proposed by Jeon and Kim (2016). Monte Carlo simulation is also conducted to compare the empirical powers of the proposed method with previous methods.

• Proceedings of the Korean Statistical Society Conference
• /
• 2005.05a
• /
• pp.73-77
• /
• 2005

본 논문은 시계열에 내재된 설${\cdot}$추석 등 음력 명절효과의 존재를 검정하기 위해 RegARIMA 모형의 잔차에 대한 t-검정 통계량을 제시하였으며 Box-plot에 의한 그래프적 진단을 시도하였다. 제시된 t-검정 결과를 X-12-ARIMA의 AICC-사전검정 및 RegARIMA 모형에 의해 추정된 명절효과 회귀계수의 t-값과 비교하였다. 사용된 명절효과 변수는 Bell과 Hillmer(1983)의 명절효과 변수이다.

• The Korean Journal of Applied Statistics
• /
• v.13 no.2
• /
• pp.407-414
• /
• 2000

Carroll and Ruppert(1988) analyzed the esterase assay data with regression model based on quasi-likelihood. Jung and Lee(1997) introduced a goodness-of-fit test for testing the adequacy of the quasi-likelihood and claimed that there is no gross inadequacy with the model because their test was not rejected. However, Lee and Xelder(199S)'s residual plots revealed that the model did not sufficiently reflect the increase of the variance with that of the mean. In this paper, we re-analyze the esterase assay data with the joint modelling of mean and dispersion in Lee and l\elder(1998) and evaluate the validity of the fitted model by applying the residual plots. And it is illustrated that Lee and Nelder(199S)'s restricted likelihood is more efficient in goodness-of-fit test for the dispersion model.

• Journal of the Korean Data and Information Science Society
• /
• v.22 no.4
• /
• pp.701-715
• /
• 2011

The object of study is what kinds of contents are preferable depending on regions and terminal types for high speed mobile internet. We consider 10 contents, 9 region2, and 2 terminal types. The methods are the adjusted residuals, the corresponding analysis, and the multiple corresponding analysis. The results are different, which comes from different mathematical models. 50% of results are same between the corresponding analysis and the multiple corresponding analysis. 26.3% are same between the adjusted residuals and the corresponding analysis. 21.1% are same between the adjusted residuals and the multiple corresponding analysis. We recommend the adjusted residuals because it can test hypothesis for preferring contents. The content which is not chosen by three methods is business.

• Journal of Korean Society of Forest Science
• /
• v.106 no.3
• /
• pp.320-329
• /
• 2017

This study aims to evaluate whether the accuracy of LiDAR DEM is affected by the changes of the five input levels ('1','3','5','7' and '9') of median parameter ($F_{md}$), mean parameter ($F_{mn}$) of the Filtering Algorithm (FA) in the GroundFilter module and median parameter ($I_{md}$), mean parameter ($I_{mn}$) of the Interpolation Algorithm (IA) in the GridSurfaceCreate module of the FUSION in order to present the combination of parameter levels producing the most accurate LiDAR DEM. The accuracy is measured by the residuals calculated by difference between the field elevation values and their corresponding DEM elevation values. A multi-way ANOVA is used to statistically examine whether there are effects of parameter level changes on the means of the residuals. The Tukey HSD is conducted as a post-hoc test. The results of the multi- way ANOVA test show that the changes in the levels of $F_{md}$, $F_{mn}$, $I_{mn}$ have significant effects on the DEM accuracy with the significant interaction effect between $F_{md}$ and $F_{mn}$. Therefore, the level of $F_{md}$, $F_{mn}$, and the interaction between two variables are considered to be factors affecting the accuracy of LiDAR DEM as well as the level of $I_{mn}$. As the results of the Tukey HSD test on the combination levels of $F_{md}{\ast}F_{mn}$, the mean of residuals of the '$9{\ast}3$' combination provides the highest accuracy while the '$1{\ast}1$' combination provides the lowest one. Regarding $I_{mn}$ levels, the mean of residuals of the both '3' and '1' provides the highest accuracy. This study can contribute to improve the accuracy of the forest attributes as well as the topographic information extracted from the LiDAR data.

• The Korean Journal of Applied Statistics
• /
• v.8 no.2
• /
• pp.187-200
• /
• 1995

This paper is concerned with a regression model with nonlinear time series errors. Testing procedures for linearity of error terms are studied. To this end, large-sample properties of estimators of regression parameters and autoregression parameter are obtained. These results are then used to develop test statistics for testing linearity of errors. Some simulation studies are shown.