• Journal of The Geomorphological Association of Korea
• /
• 제24권2호
• /
• pp.91-101
• /
• 2017

Quantitative forecasting methods based on spatial data and geographic information system have been used in predicting the landslide location. This study compared the simulated results of logistic, Bayesian, and maximum entropy models to understand the uncertainties of each model and identify the main factors that influence landslide. The study area is Boeun gun where 388 landslides occurred in the year of 1998. The verification results showed that the AUC of the three models was 0.84. However, the landslide susceptibility distribution of Maxent model was different from those of the other two models. With the same landslide occurrence data, the result of high susceptible area in Maxent model is smaller than Logistic or Bayesian. Maxent model, however, proved to be more efficient in predicting landslide than the other two models. In Maxent's simulations, the responsible factors for landslide susceptibility are timber age class, land cover, timber diameter, crown closure, and soil drainage. The results suggest that it is necessary to consider the possibility of overestimation when using Logistic or Bayesian model, and forest management around the study area can be an effective way to minimize landslide possibility.

• Journal of the Korean Data and Information Science Society
• /
• 제18권4호
• /
• pp.1123-1134
• /
• 2007

This article addresses the problem of testing whether the shape parameters in k independent Weibull populations are equal. We propose a Bayesian model selection procedure for equality of the shape parameters. The noninformative prior is usually improper which yields a calibration problem that makes the Bayes factor to be defined up to a multiplicative constant. So we propose the objective Bayesian model selection procedure based on the fractional Bayes factor and the intrinsic Bayes factor under the reference prior. Simulation study and a real example are provided.

• Proceedings of the Korean Statistical Society Conference
• /
• 한국통계학회 2003년도 춘계 학술발표회 논문집
• /
• pp.183-189
• /
• 2003

A momentum threshold autoregressive (MTAR) model, a nonlinear autoregressive model, is analyzed in a Bayesian framework. Parameter estimation in the presence of missing data is done by using Markov chain Monte Carlo methods. We also propose simple Bayesian test procedures for asymmetry and unit roots. The proposed method is applied to a set of Korea unemployment rate data and reveals evidence for asymmetry and a unit root.

• Journal of the Korean Data and Information Science Society
• /
• 제15권4호
• /
• pp.1033-1043
• /
• 2004

We consider the simultaneous Bayesian estimation for the normal means based on different noninformative type hyperpriors in hierarchical model. We provide numerical example using the famous baseball data in Efron and Morris (1975) for illustration.

• Journal of Korean Society for Quality Management
• /
• 제19권1호
• /
• pp.72-82
• /
• 1991

This study deals with the problem of water resources development by using bayesian theory. The purpose of this study is to develop the optimal decision model by applying bayesian theory which determine the optimal alternative in water resources development system. A relevant mathematical model to find an optimal solution formulated and then used in developing an efficient water resources that determine optimal alternative. A numerical example is solved to illustrate the algorithm developed.

• Communications for Statistical Applications and Methods
• /
• 제2권2호
• /
• pp.115-125
• /
• 1995

This paper deals with the Bayesian estimation for the NLAR(1) model with Laplacian marginals. Assuming the independent uniform priors for two parameters of the NLAT(1) model, the griddy Gbbs sampler by Ritter and Tanner(1992) is used to obtain the Bayesian estimates. Random numbers generated form the uniform priors ate used as the grids for each parameter. Some simulations are conducted and compared with the maximum likelihood estimation result.

• Journal of the Korean Data and Information Science Society
• /
• 제10권1호
• /
• pp.147-154
• /
• 1999

Using a noninformative prior and an inverted gamma prior, the Bayesian predictive density and the prediction intervals for a future observation or the p - th order statistic of n' future observations from the censord Pareto model have been obtained. In additions, numerical examples are given in order to illustrate the proposed predictive procedure.

• Proceedings of the Korean Statistical Society Conference
• /
• 한국통계학회 2003년도 추계 학술발표회 논문집
• /
• pp.9-14
• /
• 2003

A Bayesian model-based clustering method is proposed for clustering objects on the basis of dissimilarites. This combines two basic ideas. The first is that tile objects have latent positions in a Euclidean space, and that the observed dissimilarities are measurements of the Euclidean distances with error. The second idea is that the latent positions are generated from a mixture of multivariate normal distributions, each one corresponding to a cluster. We estimate the resulting model in a Bayesian way using Markov chain Monte Carlo. The method carries out multidimensional scaling and model-based clustering simultaneously, and yields good object configurations and good clustering results with reasonable measures of clustering uncertainties. In the examples we studied, the clustering results based on low-dimensional configurations were almost as good as those based on high-dimensional ones. Thus tile method can be used as a tool for dimension reduction when clustering high-dimensional objects, which may be useful especially for visual inspection of clusters. We also propose a Bayesian criterion for choosing the dimension of the object configuration and the number of clusters simultaneously. This is easy to compute and works reasonably well in simulations and real examples.

• Communications for Statistical Applications and Methods
• /
• 제6권3호
• /
• pp.801-812
• /
• 1999

Change-point hazard rate models arise for example in applying "burn-in" techniques to screen defective items and in studing times until undesirable side effects occur in clinical trials. Sometimes in screening defectives it might be sensible to model two stages of burn-in. In a clinical trial there might be an initial hazard rate for a side effect which after a period of time changes to an intermediate hazard rate before settling into a long term hazard rate. In this paper we consider the multiple change points hazard rate model. The classical approach's asymptotics can be poor for the small to all moderate sample sizes often encountered in practice. We propose a Bayesian approach avoiding asymptotics to provide more reliable inference conditional only upon the data actually observed. The Bayesian models can be fitted using simulation methods. Model comparison is made using recently developed Bayesian model selection criteria. The above methodology is applied to a generated data and to a generated data and the Lawless(1982) failure times of electrical insulation.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제13권9호
• /
• pp.4684-4705
• /
• 2019

Collaborative filtering recommender systems are vulnerable to shilling attacks in which malicious users may inject biased profiles to promote or demote a particular item being recommended. To tackle this problem, many robust collaborative recommendation methods have been presented. Unfortunately, the robustness of most methods is improved at the expense of prediction accuracy. In this paper, we construct a robust Bayesian probabilistic matrix factorization model for collaborative filtering recommender systems by incorporating the detection of user anomaly rating behaviors. We first detect the anomaly rating behaviors of users by the modified K-means algorithm and target item identification method to generate an indicator matrix of attack users. Then we incorporate the indicator matrix of attack users to construct a robust Bayesian probabilistic matrix factorization model and based on which a robust collaborative recommendation algorithm is devised. The experimental results on the MovieLens and Netflix datasets show that our model can significantly improve the robustness and recommendation accuracy compared with three baseline methods.