• Korean Journal of Remote Sensing
• /
• v.39 no.5_4
• /
• pp.1125-1134
• /
• 2023

Out of the total 17,000 reservoirs in Korea, 13,600 small agricultural reservoirs do not have hydrological measurement facilities, making it difficult to predict water storage volume and appropriate operation. This paper examined univariate and multivariate long short-term memory (LSTM) modeling to predict the storage rate of agricultural reservoirs using remote sensing and artificial intelligence. The univariate LSTM model used only water storage rate as an explanatory variable, and the multivariate LSTM model added n-day accumulative precipitation and date of year (DOY) as explanatory variables. They were trained using eight years data (2013 to 2020) for Idong Reservoir, and the predictions of the daily water storage in 2021 were validated for accuracy assessment. The univariate showed the root-mean square error (RMSE) of 1.04%, 2.52%, and 4.18% for the one, three, and five-day predictions. The multivariate model showed the RMSE 0.98%, 1.95%, and 2.76% for the one, three, and five-day predictions. In addition to the time-series storage rate, DOY and daily and 5-day cumulative precipitation variables were more significant than others for the daily model, which means that the temporal range of the impacts of precipitation on the everyday water storage rate was approximately five days.

• Communications for Statistical Applications and Methods
• /
• v.17 no.5
• /
• pp.697-712
• /
• 2010

This paper is concerned with the statistical analysis and development of stochastic models for the demand for life insurance policy loans. For these, firstly the characteristics of the regression trend, periodicity and dependence of the monthly demand for life insurance policy loans are investigated by a statistical analysis of the monthly demand data for the years 1999 through 2008. Secondly, the causal relationships between the demand for life insurance policy loans and the economic variables including unemployment rate and inflation rate for the period are investigated. The results show that inflation rate is main factor influencing policy loan demands. The overall evidence, however, failed to establish unidirectional causality relationships between the demand series and the other variables under study. Finally, based on these, univariate time series model and transfer function model where the demand series is related to one input series are derived, respectively, for the prediction of the demand for life insurance policy loans. A statistical procedure for using the model to predict the demand for life insurance policy loans is also proposed.

• Journal of Korea Water Resources Association
• /
• v.42 no.9
• /
• pp.725-736
• /
• 2009

A flood event can be defined by three characteristics; peak discharge, total flood volume, and flood duration, which are correlated each other. However, a conventional flood frequency analysis for the hydrological plan, design, and operation has focused on evaluating only the amount of peak discharge. The interpretation of this univariate flood frequency analysis has a limitation in describing the complex probability behavior of flood events. This study proposed a bivariate flood frequency analysis using a Gumbel mixed model for the flood evaluation. A time series of annual flood events was extracted from observations of inflow to the Soyang River Dam and the Daechung Dam, respectively. The joint probability distribution and return period were derived from the relationship between the amount of peak discharge and the total volume of flood runoff. The applicability of the Gumbel mixed model was tested by comparing the return periods acquired from the proposed bivariate analysis and the conventional univariate analysis.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.26 no.3B
• /
• pp.279-289
• /
• 2006

The reconstruction of low dimension nonlinear behavior from the hydrologic time series has been an active area of research in the last decade. In this study, we present the applications of a powerful state space reconstruction methodology using the method of Support Vector Machines (SVM) to the Great Salt Lake (GSL) volume. SVMs are machine learning systems that use a hypothesis space of linear functions in a Kernel induced higher dimensional feature space. SVMs are optimized by minimizing a bound on a generalized error (risk) measure, rather than just the mean square error over a training set. The utility of this SVM regression approach is demonstrated through applications to the short term forecasts of the biweekly GSL volume. The SVM based reconstruction is used to develop time series forecasts for multiple lead times ranging from the period of two weeks to several months. The reliability of the algorithm in learning and forecasting the dynamics is tested using split sample sensitivity analyses, with a particular interest in forecasting extreme states. Unlike previously reported methodologies, SVMs are able to extract the dynamics using only a few past observed data points (Support Vectors, SV) out of the training examples. Considering statistical measures, the prediction model based on SVM demonstrated encouraging and promising results in a short-term prediction. Thus, the SVM method presented in this study suggests a competitive methodology for the forecast of hydrologic time series.

• Advances in environmental research
• /
• v.4 no.4
• /
• pp.233-246
• /
• 2015

High amounts of air pollution in crowded urban areas are always considered as one of the major environmental challenges especially in developing countries. Despite the errors in air pollution prediction, the forecasting of future data helps air quality management make decisions promptly and properly. We studied the air quality of the Aqdasiyeh location in Tehran using factor analysis and the Box-Jenkins time series methods. The Air Quality Control Company (AQCC) of the Municipality of Tehran monitors seven daily air quality parameters, including carbon monoxide (CO), Nitrogen Monoxide (NO), Nitrogen dioxide ($NO_2$), $NO_x$, ozone ($O_3$), particulate matter ($PM_{10}$) and sulfur dioxide ($SO_2$). We applied the AQCC data for our study. According to the results of the factor analysis, the air quality parameters were divided into two factors. The first factor included CO, $NO_2$, NO, $NO_x$, and $O_3$, and the second was $SO_2$ and $PM_{10}$. Subsequently, the Box- Jenkins time series was applied to the two mentioned factors. The results of the statistical testing and comparison of the factor data with the predicted data indicated Auto Regressive Integrated Moving Average (0, 0, 1) was appropriate for the first factor, and ARIMA (1, 0, 1) was proper for the second one. The coefficient of determination between the factor data and the predicted data for both models were 0.98 and 0.983 which may indicate the accuracy of the models. The application of these methods could be beneficial for the reduction of developing numbers of mathematical modeling.

• The Korean Journal of Applied Statistics
• /
• v.27 no.6
• /
• pp.1049-1068
• /
• 2014

Autoregressive models are used to analyze an univariate time series data; however, these methods can be inappropriate when a structural break appears in a time series since they assume that a trend is consistent. Threshold autoregressive models (popular regime-switching models) have been proposed to address this problem. Recently, the models have been extended to two regime-switching models with delay parameter. We discuss two regime-switching threshold autoregressive models from a Bayesian point of view. For a Bayesian analysis, we consider a parametric threshold autoregressive model and a nonparametric threshold autoregressive model using Dirichlet process prior. The posterior distributions are derived and the posterior inferences is performed via Markov chain Monte Carlo method and based on two Bayesian threshold autoregressive models. We present a simulation study to compare the performance of the models. We also apply models to gross domestic product data of U.S.A and South Korea.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.30 no.3B
• /
• pp.257-267
• /
• 2010

A flood event can be characterized by three attributes such as peak discharge, total flood volume, and flood duration, which are correlated each other. However, the amount of peak discharge is only used to evaluate the flood events for the hydrological plan and design. The univariate analysis has a limitation in describing the complex probability behavior of flood events. Thus, the univariate analysis cannot derive satisfying results in flood frequency analysis. This study proposed bivariate flood frequency analysis methods for evaluating flood events considering correlations among attributes of flood events. Parametric distributions such as Gumbel mixed model and bivariate gamma distribution, and a non-parametric model using a bivariate kernel function were introduced in this study. A time series of annual flood events were extracted from observations of inflow to the Soyang River Dam and the Daechung Dam, respectively. The joint probability distributions and return periods were derived from the relationship between the amount of peak discharge and the total volume of flood runoff. Applicabilities of bivariate flood frequency analysis were examined by comparing the return period acquired from the proposed bivariate analyses and the conventional univariate analysis.

• Atmosphere
• /
• v.19 no.2
• /
• pp.199-211
• /
• 2009

In this paper, we examine the multiple change-point and change pattern in the 54 years (1954-2007) time series of the annual and the heavy precipitation characteristics (amount, days and intensity) averaged over South Korea. A Bayesian approach is used for detecting of mean and/or variance changes in a sequence of independent univariate normal observations. Using non-informative priors for the parameters, the Bayesian model selection is performed by the posterior probability through the intrinsic Bayes factor of Berger and Pericchi (1996). To investigate the significance of the changes in the precipitation characteristics between before and after the change-point, the posterior probability and 90% highest posterior density credible intervals are examined. The results showed that no significant changes have occurred in the annual precipitation characteristics (amount, days and intensity) and the heavy precipitation intensity. On the other hand, a statistically significant single change has occurred around 1996 or 1997 in the heavy precipitation days and amount. The heavy precipitation amount and days have increased after the change-point but no changes in the variances.

• International Journal of Computer Science & Network Security
• /
• v.23 no.6
• /
• pp.127-139
• /
• 2023

Numerous genuine issues, for example, financial exchange expectation, climate determining and so forth has inalienable arbitrariness related with them. Receiving a probabilistic system for forecast can oblige this dubious connection among past and future. Commonly the interest is in the contingent likelihood thickness of the arbitrary variable included. One methodology for expectation is with time arrangement and auto relapse models. In this work, liner expectation technique and approach for computation of forecast coefficient are given and likelihood of blunder for various assessors is determined. The current methods all need in some regard assessing a boundary of some accepted arrangement. In this way, an elective methodology is proposed. The elective methodology is to gauge the restrictive thickness of the irregular variable included. The methodology proposed in this theory includes assessing the (discretized) restrictive thickness utilizing a Markovian definition when two arbitrary factors are genuinely needy, knowing the estimation of one of them allows us to improve gauge of the estimation of the other one. The restrictive thickness is assessed as the proportion of the two dimensional joint thickness to the one-dimensional thickness of irregular variable at whatever point the later is positive. Markov models are utilized in the issues of settling on an arrangement of choices and issue that have an innate transience that comprises of an interaction that unfurls on schedule on schedule. In the nonstop time Markov chain models the time stretches between two successive changes may likewise be a ceaseless irregular variable. The Markovian methodology is especially basic and quick for practically all classes of classes of issues requiring the assessment of contingent densities.

• Journal of Korean Neurosurgical Society
• /
• v.48 no.3
• /
• pp.230-235
• /
• 2010

Objective : The aim of this study is to compare the surgical outcome of the initial and recent surgical cases, during our 15-years experience, in terms of the surgical strategies and the prognostic factors for surgically remediable epilepsy. Methods : We retrospectively reviewed and compared the surgical outcomes between the initial 256 (Group I) and recent 139 (Group II) patients according to the time period of operation for a total of 518 consecutive epilepsy surgeries at our institution since 1992. The patients of the middle intermediate period, which were subjected to changed surgical strategies, were excluded. Results : The surgical outcome data from the initial and recent groups showed a much improved outcome for patients who underwent temporal lobe epilepsy (TLE) surgery over time. The number of patients with a good outcome (Engel class I-II) was much increased from 87.7% (178 TLE cases of Group I) to 94.8% (79 TLE cases of Group II) and this was statistically significant (p = 0.0324) on univariate analysis. Other remarkable changes were the decreased performance of intracranial invasive studies from 43.5% in Group I to 30.9% in Group II due to the advanced neuroimaging tools. The strip/grid ratio was reduced from 131/32 in Group I to 17/25 in Group II, because of a markedly reduced mesial TLE surgery and an increased extratemporal epilepsy surgery. Conclusion : Our results show that surgical outcome of epilepsy surgery has improved over time and it has shown to be efficient to control medically intractable epilepsy. Appropriate patient selection, comprehensive preoperative assessments and more extensive resection are associated with good postoperative outcomes.