• Journal of Information Processing Systems
• /
• v.15 no.3
• /
• pp.694-706
• /
• 2019

In this study, we applied the long short-term memory (LSTM) model to classify the cryptocurrency price time series. We collected historic cryptocurrency price time series data and preprocessed them in order to make them clean for use as train and target data. After such preprocessing, the price time series data were systematically encoded into the three-dimensional price tensor representing the past price changes of cryptocurrencies. We also presented our LSTM model structure as well as how to use such price tensor as input data of the LSTM model. In particular, a grid search-based k-fold cross-validation technique was applied to find the most suitable LSTM model parameters. Lastly, through the comparison of the f1-score values, our study showed that the LSTM model outperforms the gradient boosting model, a general machine learning model known to have relatively good prediction performance, for the time series classification of the cryptocurrency price trend. With the LSTM model, we got a performance improvement of about 7% compared to using the GB model.

• Communications for Statistical Applications and Methods
• /
• v.29 no.3
• /
• pp.353-371
• /
• 2022

This paper analyzes death counts after World War II of several countries to identify and to compare their stochastic structures. The stochastic structures that this paper entertains are three structural time series models, a local level with a random walk model, a fixed local linear trend model and a local linear trend model. The structural time series models assume that a time series can be formulated directly with the unobserved components such as trend, slope, seasonal, cycle and daily effect. Random effect of each unobserved component is characterized by its own stochastic structure and a distribution of its irregular component. The structural time series models use the Kalman filter to estimate unknown parameters of a stochastic model, to predict future data, and to do filtering data. This paper identifies the best-fitted stochastic model for three types of death counts (Female, Male and Total) of each country. Two diagnostic procedures are used to check the validity of fitted models. Three criteria, AIC, BIC and SSPE are used to select the best-fitted valid stochastic model for each type of death counts of each country.

• Communications for Statistical Applications and Methods
• /
• v.12 no.2
• /
• pp.359-364
• /
• 2005

This paper considers a specification test of conditional Poisson regression model for time series count data. Although conditional models for count data have received attention and proposed in several ways, few studies focused on checking its adequacy. Motivated by the test of martingale difference assumption, a specification test via Ljung-Box statistic is proposed in the conditional model of the time series count data. In order to illustrate the performance of Ljung- Box test, simulation results will be provided.

• Communications for Statistical Applications and Methods
• /
• v.18 no.1
• /
• pp.79-87
• /
• 2011

A spatial time series model was used for analyzing the method of spatial time series (not the ARIMA model that is popular for analyzing spatial time series) by using chicken pox data which is a highly contagious disease and grid data due to ARIMA not reflecting the spatial processes. Time series model contains a weighting matrix, because that spatial time series model influences the time variation as well as the spatial location. The weighting matrix reflects that the more geographically contiguous region has the higher spatial dependence. It is hypothesized that the weighting matrix gives neighboring areas the same influence in the study of the spatial time series model. Therefore, we try to present the conclusion with a weighting matrix in a way that gives the same weight to existing neighboring areas in the study of the suitability of the STARMA model, spatial time series model and STBL model, in the comparative study of the predictive power for statistical inference, and the results. Furthermore, through the Kalman-Filter method we try to show the superiority of the Kalman-Filter method through a parameter assumption and the processes of prediction.

• The Korean Journal of Applied Statistics
• /
• v.33 no.2
• /
• pp.147-160
• /
• 2020

In this paper, we study an algorithm that automatically determines the orders of past observations and conditional mean values that play an important role in count time series models. Based on the orders of the ARIMA model, the algorithm constitutes the order candidates group for time series generalized linear models and selects the final model based on information criterion among the combinations of the order candidates group. To evaluate the proposed algorithm, we perform small simulations and empirical analysis according to underlying models and time series as well as compare forecasting performances with the ARIMA model. The results of the comparison confirm that the time series generalized linear model offers better performance than the ARIMA model for the count time series analysis. In addition, the empirical analysis shows better performance in mid and long term forecasting than the ARIMA model.

• Journal of applied mathematics & informatics
• /
• v.5 no.2
• /
• pp.277-284
• /
• 1998

The reliability and an estimate for it are derived for series-parallel and parallel-deries stress-strength model under assumption that all components are subjected to a common stress. We also obtain the asymptotic normal distribution of the estimate.

• Journal of Korean Society of Water and Wastewater
• /
• v.21 no.6
• /
• pp.701-707
• /
• 2007

The information on the incoming load to wastewater treatment plants is not often available to apply to evaluate effects of control actions on the field plant. In this study, a time series model was developed to forecast influent flow rate, BOD, COD, SS, TN and TP concentrations using field operating data. The developed time series model could predict 1 day ahead forecasting results accurately. The coefficient of determination between measured data and 1 day ahead forecasting results has a range from 0.8898 to 0.9971. So, the corelation is relatively high. We made forecasting program based on the time series model developed and hope that the program will assist the operators in the stable operation in wastewater treatment plants.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.66 no.3
• /
• pp.493-499
• /
• 2017

This paper proposes a combined very-short-term and short-term multi-variate time-series model for 24 hour load forecasting. First, the best model for very-short-term and short-term load forecasting is selected by considering the least error value, and then they are combined by the optimal forecasting time. The actual load data of industry complex is used to show the effectiveness of the proposed model. As a result the load forecasting accuracy of the combined model has increased more than a single model for 24 hour load forecasting.

• Journal of Industrial Convergence
• /
• v.2 no.2
• /
• pp.127-147
• /
• 2004

This paper examines the time-series relations among expected return, risk, and book-to-market(B/M) at the portfolio level. The time-series analysis is a natural alternative to cross-sectional regressions. An alternative feature of the time-series regressions is that they focus on changes in expected returns, not on average returns. Using the time-series analysis, we can directly test whether the three-factor model explains time-varying expected returns better than the characteristic-based model. These results should help distinguish between the risk and mispricing stories. We find that B/M is strongly associated with changes in risk, as measured by the Fama and French(1993) three-factor model. After controlling for changes in risk, B/M contains little additional information about expected returns. The evidence suggests that the three-factor model explains time-varying expected returns better than the characteristic-based model.

• Asia-Pacific Journal of Atmospheric Sciences
• /
• v.54 no.4
• /
• pp.611-622
• /
• 2018

Observed climate data are processed under the assumption that their time series are stationary, as in multi-step temperature and precipitation prediction, which usually leads to low prediction accuracy. If a climate system model is based on a single prediction model, the prediction results contain significant uncertainty. In order to overcome this drawback, this study uses a method that integrates ensemble prediction and a stepwise regression model based on a mean-valued generation function. In addition, it utilizes empirical mode decomposition (EMD), which is a new method of handling time series. First, a non-stationary time series is decomposed into a series of intrinsic mode functions (IMFs), which are stationary and multi-scale. Then, a different prediction model is constructed for each component of the IMF using numerical ensemble prediction combined with stepwise regression analysis. Finally, the results are fit to a linear regression model, and a short-term climate prediction system is established using the Visual Studio development platform. The model is validated using temperature data from February 1957 to 2005 from 88 weather stations in Guangxi, China. The results show that compared to single-model prediction methods, the EMD and ensemble prediction model is more effective for forecasting climate change and abrupt climate shifts when using historical data for multi-step prediction.