• KIPS Transactions on Computer and Communication Systems
• /
• v.8 no.4
• /
• pp.79-86
• /
• 2019

This paper describes a time-series data prediction based on artificial neural networks (ANN). In this study, a batch based ANN model and a stochastic ANN model have been implemented using TensorFlow libraries. Each model are evaluated by comparing training and testing errors that are measured through experiment. To train and test each model, tax dataset was used that are collected from the government website of indiana state budget agency in USA from 2001 to 2018. The dataset includes tax incomes of individual, product sales, company, and total tax incomes. The experimental results show that batch model reveals better performance than stochastic model. Using the batch scheme, we have conducted a prediction experiment. In the experiment, total taxes are predicted during next seven months, and compared with actual collected total taxes. The results shows that predicted data are almost same with the actual data.

• Journal of Korea Multimedia Society
• /
• v.25 no.1
• /
• pp.82-91
• /
• 2022

In this paper, time series prediction method of photovoltaic power is introduced using seasonal autoregressive integrated moving average (SARIMA). In order to obtain the best fitting model by a time series method in the absence of an environmental sensor, this research was used data below 50% of cloud cover. Three samples were extracted by time intervals from the raw data. After that, the best fitting models were derived from mean absolute percentage error (MAPE) with the minimum akaike information criterion (AIC) or beysian information criterion (BIC). They are SARIMA (1,0,0)(0,2,2)14, SARIMA (1,0,0)(0,2,2)28, SARIMA (2,0,3)(1,2,2)55. Generally parameter of model derived from BIC was lower than AIC. SARIMA (2,0,3)(1,2,2)55, unlike other models, was drawn by AIC. And the performance of models obtained by SARIMA was compared. MAPE value was affected by the seasonal period of the sample. It is estimated that long seasonal period samples include atmosphere irregularity. Consequently using 1 hour or 30 minutes interval sample is able to be helpful for prediction accuracy improvement.

• Knowledge Management Research
• /
• v.22 no.4
• /
• pp.119-133
• /
• 2021

As a large amount of data is produced in each industry, a number of time series pattern prediction studies are being conducted to make quick business decisions. However, there is a limit to predicting specific patterns in nonlinear time series data due to the uncertainty inherent in the data, and there are difficulties in making strategic decisions in corporate management. In addition, in recent decades, various studies have been conducted on data such as demand/supply and financial markets that are suitable for industrial purposes to predict time series data of irregular random walk models, but predict specific rules and achieve sustainable corporate objectives There are difficulties. In this study, the prediction results were compared and analyzed using the Chaos analysis method for roulette data and financial market data, and meaningful results were derived. And, this study confirmed that chaos analysis is useful for finding a new method in analyzing time series data. By comparing and analyzing the characteristics of roulette games with the time series of Korean stock index future, it was derived that predictive power can be improved if the trend is confirmed, and it is meaningful in determining whether nonlinear time series data with high uncertainty have a specific pattern.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.05a
• /
• pp.35-37
• /
• 2021

Analyzing and predicting foreign tourists' demand is a crucial research topic in the tourism industry because it profoundly influences establishing and planning tourism policies. Since foreign tourist data is influenced by various external factors, it has a characteristic that there are many subtle changes over time. Therefore, in recent years, research is being conducted to design a prediction model by reflecting various external factors such as economic variables to predict the demand for tourists inbound. However, the regression analysis model and the recurrent neural network model, mainly used for time series prediction, did not show good performance in time series prediction reflecting various variables. Therefore, we design a foreign tourist demand prediction model that complements these limitations using a convolutional neural network. In this paper, we propose a model that predicts foreign tourists' demand by designing a one-dimensional convolutional neural network that reflects foreign tourist data for the past ten years provided by the Korea Tourism Organization and additionally collected external factors as input variables.

• 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.

• Journal of Computing Science and Engineering
• /
• v.5 no.4
• /
• pp.338-345
• /
• 2011

Nowadays mobile devices are used for various applications such as making voice/video calls, browsing the Internet, listening to music etc. The average battery consumption of each of these activities and the length of time a user spends on each one determines the battery lifetime of a mobile device. Previous methods have provided predictions of battery lifetime using a static battery consumption rate that does not consider user characteristics. This paper proposes an approach to predict a mobile device's available battery lifetime based on usage patterns. Because every user has a different pattern of voice calls, data communication, and video call usage, we can use such usage patterns for personalized prediction of battery lifetime. Firstly, we define one or more states that affect battery consumption. Then, we record time-series log data related to battery consumption and the use time of each state. We calculate the average battery consumption rate for each state and determine the usage pattern based on the time-series data. Finally, we predict the available battery time based on the average battery consumption rate for each state and the usage pattern. We also present the experimental trials used to validate our approach in the real world.

• Journal of the Korean Data and Information Science Society
• /
• v.23 no.6
• /
• pp.1093-1102
• /
• 2012

In this paper, we predicted the interest spread using the VAR (vector autoregressive) model. Variables used in the VAR model were selected among 56 domestic and foreign macroeconomic time series through crosscorrelation and Granger causality test. The performance of the VAR model was compared with the univariate time series model, AR (autoregressive) model, in view of MAPE (mean absolute percentage error) and RMSE (root mean square error) of forecasts for the last twelve months.

• Journal of Environmental Science International
• /
• v.9 no.1
• /
• pp.19-25
• /
• 2000

As the tideland reclamation is done on a large scale these days, construction work is active in the coastal areas. Facilities in the coastal areas must be built with the tide characteristics taken into consideration. Thus the tide characteristics affect the overall reclamation plan. The analysis of the tide data boils down to a harmonic analysis of the hourly changes of long-term tide data and extraction of unharmonic coefficients from the results. Since considerable amount of tide data of the West Coast are available, the existing data can be collected and can be used to obtain the temporal changes of the tide by being fitted into the tide prediction model. The goal of this thesis lies in assessing whether the mean sea level used in the field agrees with the analysis results from the long-term observation data obtained with their homogeneity guaranteed. To achieve this goal, the research was conducted as follows. First the present conditions of the observation stations, the land level standard, and the sea level standard were analyzed to set up a time series model formula for representing them. To secure the homogeneity of the time series, each component was separated. Lastly the mean sea level used in the field was assessed based on the results obtained form the analysis of the time series.

• Journal of the Korea Society of Computer and Information
• /
• v.16 no.12
• /
• pp.11-21
• /
• 2011

An embedded sensor is significantly influenced by its spatial environment, such as barriers or distance, through low power and signal strength. Due to these causes, noise data frequently occur in an embedded sensor. Because the information acquired from the embedded sensor exists in a time series, it is hard to detect an error which continuously takes place in the time series information on a realtime basis. In this paper, we proposes an error detection method based on time-series prediction that detects error signals of embedded sensors in real time in consideration of the physical characteristics of embedded devices. The error detection method based on time-series prediction proposed in this paper determines errors in generated embedded device signals using a stable distance function. When detecting errors by monitoring signals from an embedded device, the stable distance function can detect error signals effectively by applying error weight to the latest signals. When detecting errors by monitoring signals from an embedded device, the stable distance function can detect error signals effectively by applying error weight to the latest signals.

• Water for future
• /
• v.28 no.1
• /
• pp.81-90
• /
• 1995

This study attempts to develop a stochastic system model for extension and prediction of monthly runoff series in river basins where the observed runoff data are insufficient although there are long-term hydrometeorological records. For this purpose, univariate models of a seasonal ARIMA type are derived from the time series analysis of monthly runoff, monthly precipitation and monthly evaporation data with trend and periodicity. Also, a causual model of multiple input-single output relationship that take monthly precipitation and monthly evaporation as input variables-monthly runoff as output variable is built by the cross-correlation analysis of each series. The performance of the univariate model and the multiple input-output model were examined through comparisons between the historical and the generated monthly runoff series. The results reveals that the multiple input-output model leads to the improved accuracy and wide range of applicability when extension and prediction of monthly runoff series is required.