• The KIPS Transactions:PartB
• /
• v.9B no.3
• /
• pp.277-286
• /
• 2002

Genetic programming (GP) has been combined with quantum mechanical perturbation theory to make a new algorithm to construct mathematical models and perform predictions for chaotic time series from real world. Procedural similarities between time series modeling and perturbation theory to solve quantum mechanical wave equations are discussed, and the exemplary GP approach for implementing them is proposed. The approach is based on multiple populations and uses orthogonal functions for GP function set. GP is applied to original time series to get the first mathematical model. Numerical values of the model are subtracted from the original time series data to form a residual time series which is again subject to GP modeling procedure. The process is repeated until predetermined terminating conditions are met. The algorithm has been successfully applied to construct highly effective mathematical models for many real world chaotic time series. Comparisons with other methodologies and topics for further study are also introduced.

• Journal of Korean Society of Transportation
• /
• v.26 no.1
• /
• pp.51-63
• /
• 2008

Jeju-Mainland demand for air passenger is variated by the season because most of the demands stem from the leisure travel. This research is to estimate the econometrics demand models(A simple time series model and the partial adjustment model) and elasticities of each models for the Jeju-Mainland domestic routes air travel market using the time series aggregate data between the year 1996 and 2005. As the result of estimating, income elasticity was evaluated to be elastic(1.55) and fare elasticity was inelastic(-0.49${\sim}$-0.59) for A simple time series models. In the partial adjustment model's case, income elasticity was evaluated to be inelastic(0.51) in short-run whereas it was evaluated to be elastic(1.88) in long-run. Fare elasticity was evaluated to be inelastic in short-run(high-demand season: -0.13, slack season: -0.20) and long-run(high-demand season: -0.48, slack season: -0.72).

• Journal of Korean Society of Transportation
• /
• v.29 no.4
• /
• pp.95-101
• /
• 2011

This study proposes the accident estimation model developed based on the time-series cross-sectional data at 50 intersections in Cheongju. The data were collected repeatedly and accumulated from 2004 to 2007. This study focused on deriving the optimal among the various models including TSCSREG(Time Series Cross Section Regression). Four different models utilizing various elements affecting accidents were developed. Through a statistical test, it was found that the t values of independent variables of the fixed effect models were less than those of the random effect models. Two variables were then found to be positive to the accidents: the number of crosswalks at an intersection and the number of intersections.

• The Korean Journal of Applied Statistics
• /
• v.19 no.1
• /
• pp.33-41
• /
• 2006

In this paper, we analyse the volatilities in financial data such as stock prices and exchange rates in term of a class of nonlinear time series models. We compare the performance of Generalized Autoregressive Conditional Heteroscadastic(GARCH) , Integrated GARCH(IGARCH), Exponential GARCH(EGARCH) models by KOSPI (Korean stock Prices Index) data. The estimation for the parameters in the models was carried out by the ML methods.

• Journal of the Korean Data and Information Science Society
• /
• v.22 no.5
• /
• pp.829-837
• /
• 2011

Since Harvey (1989), many approaches for applying unobserved components (UC) models to both univariate and multivariate time series analysis have been developed. However, practitioners still tend to use traditional methods such as exponential smoothing or ARIMA models for modeling and predicting time series data. It is well known that the UC model combines the flexibility of ARIMA models and the easy interpretability of exponential smoothing models by using unobserved components such as trend, cycle, season, and irregular components. This study reviews the UC model and compares its relative performances with those of the other models in modeling and predicting the real gross domestic products (GDP) in Korea. We conclude that the optimal model is the UC model on basis of root mean squared error.

• Journal of the Korea Society of Computer and Information
• /
• v.27 no.5
• /
• pp.55-67
• /
• 2022

Sensor data can provide fault diagnosis for equipment. However, the cause analysis for fault results of equipment is not often provided. In this study, we propose an explainable convolutional neural network framework for the sensor-based time series classification model. We used sensor-based time series dataset, acquired from vehicles equipped with sensors, and the Wafer dataset, acquired from manufacturing process. Moreover, we used Cycle Signal dataset, acquired from real world mechanical equipment, and for Data augmentation methods, scaling and jittering were used to train our deep learning models. In addition, our proposed classification models are convolutional neural network based models, FCN, 1D-CNN, and ResNet, to compare evaluations for each model. Our experimental results show that the ResNet provides promising results in the context of time series classification with accuracy and F1 Score reaching 95%, improved by 3% compared to the previous study. Furthermore, we propose XAI methods, Class Activation Map and Layer Visualization, to interpret the experiment result. XAI methods can visualize the time series interval that shows important factors for sensor data classification.

• The Journal of Bigdata
• /
• v.8 no.2
• /
• pp.15-25
• /
• 2023

Extreme learning machines (ELMs) are a major analytical method in various prediction fields. ELMs can accurately predict even if the data contains noise or is nonlinear by learning the complex patterns of time series data through optimal learning. This study presents the recent trends of machine learning models that are mainly studied as tools for analyzing online time series data, along with the application characteristics using existing algorithms. In order to efficiently learn large-scale online data that is continuously and explosively generated, it is necessary to have a learning technology that can perform well even in properties that can evolve in various ways. Therefore, this study examines a comprehensive overview of the latest machine learning models applied to big data in the field of time series prediction, discusses the general characteristics of the latest models that learn online data, which is one of the major challenges of machine learning for big data, and how efficiently they can learn and use online time series data for prediction, and proposes alternatives.

• The Transactions of the Korean Institute of Electrical Engineers A
• /
• v.55 no.7
• /
• pp.273-280
• /
• 2006

This paper suggests the method that forecasts Dynamic Line Rating (DLR). Thermal Overload Risk Probability (TORP) of the next time is forecasted based on the present weather conditions and DLR value by Monte Carlo Simulation (MCS). To model weather elements of transmission line for MCS process, this paper will propose the use of statistical weather models that time series is applied. Also, through the case study, it is confirmed that the forecasted TORP can be utilized as a criterion that decides DLR of next time. In short, proposed method may be used usefully to keep security and reliability of transmission line by forecasting transmission capacity of the next time.

• Advances in environmental research
• /
• v.4 no.2
• /
• pp.83-104
• /
• 2015

Air Quality Index (AQI) is a pointer to broadcast short term air quality. This paper presents one day ahead AQI forecasting on seasonal basis for three major cities in Maharashtra State, India by using Artificial Neural Networks (ANN) and Genetic Programming (GP). The meteorological observations & previous AQI from 2005-2008 are used to predict next day's AQI. It was observed that GP captures the phenomenon better than ANN and could also follow the peak values better than ANN. The overall performance of GP seems better as compared to ANN. Stochastic nature of the input parameters and the possibility of auto-correlation might have introduced time lag and subsequent errors in predictions. Spectral Analysis (SA) was used for characterization of the error introduced. Correlational dependency (serial dependency) was calculated for all 24 models prepared on seasonal basis. Particular lags (k) in all the models were removed by differencing the series, that is converting each i'th element of the series into its difference from the (i-k)"th element. New time series is generated for all seasonal models in synchronization with the original time line & evaluated using ANN and GP. The statistical analysis and comparison of GP and ANN models has been done. We have proposed a promising approach of use of GP coupled with SA for real time prediction of seasonal multicity AQI.

• Communications for Statistical Applications and Methods
• /
• v.16 no.2
• /
• pp.335-348
• /
• 2009

Fuzzy regression is used as a complement or an alternative to represent the relation between variables among the forecasting models especially when the data is insufficient to evaluate the relation. Such phenomenon often occurs in seasonal time series data which require large amount of data to describe the underlying pattern. Semiparametric model is useful tool in the case where domain knowledge exists about the function to be estimated or emphasis is put onto understandability of the model. In this paper we propose fuzzy semiparametric support vector regression so that it can provide good performance on forecasting of the seasonal time series by incorporating into fuzzy support vector regression the basis functions which indicate the seasonal variation of time series. In order to indicate the performance of this method, we present two examples of predicting the seasonal time series. Experimental results show that the proposed method is very attractive for the seasonal time series in fuzzy environments.