• 한국통계학회:학술대회논문집
• /
• 한국통계학회 2005년도 추계 학술발표회 논문집
• /
• pp.43-48
• /
• 2005

We consider the problem of testing for parameter changes in time series models based on a cusum test. Although the test procedure is well-established for the mean and variance in time series models, a general parameter case has not been discussed in the literature. Therefore, here we develop a cusum test for parameter change in a more general framework. As an example, we consider the change of the parameters in an RCA(1) model and that of the autocovariances of a linear process. We also consider the variance change test for unstable models with unit roots and GARCH models.

• 전력전자학회논문지
• /
• 제26권1호
• /
• pp.32-37
• /
• 2021

This study proposes a DC series arc fault detector using a frequency analysis method called the discrete wavelet transform (DWT), in which the processing speed of the DWT algorithm is improved effectively. The processing time can be shortened because of the time characteristic of the DWT result. The performance of the developed DC series arc fault detector for a large photovoltaic system is verified with various DC series arc generation conditions. Successful DC series arc detection and improved calculation time were both demonstrated through the measured actual arc experimental result.

• 대한토목학회논문집
• /
• 제43권3호
• /
• pp.363-370
• /
• 2023

COVID-19 (Coronavirus disease 2019) 확산으로 2020년 초부터 도시철도 등 대중교통수단의 이용량이 크게 변동하였다. 이에 본 연구에서는 COVID-19 이전과 COVID-19 확산 이후, 3년 동안 도시철도 역별 일별 시계열 자료를 수집하여 DTW (Dynamic Time Warping) 거리법을 통해 시계열 군집분석 유사도를 평가하여 군집 별 회귀 중앙치를 도출하고, COVID-19 등 여러 외부 사건이 이용객 수의 변동에 미치는 영향을 시계열 충격 탐지 함수(Outlier Detection)로 진단하였다. 또한 도시철도 역의 군집 별 이용 특성을 분석하고 또한 외부 충격에 따른 승객량의 변동을 파악하였다. 향후 COVID-19 재확산 시 이용량의 유지와 회복에 대한 방안을 검토하는 데 목적을 두었다.

• Journal of Information Processing Systems
• /
• 제15권3호
• /
• 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.

• 국제학술발표논문집
• /
• The 4th International Conference on Construction Engineering and Project Management Organized by the University of New South Wales
• /
• pp.1-6
• /
• 2011

As large-scale building projects have recently increased for the residential, commercial and office facilities, construction costs for these projects have become a matter of great concern, due to their significant construction cost implications, as well as unpredictable market conditions and fluctuations in the rate of inflation during the projects' long-term construction periods. In particular, recent volatile fluctuations of construction material prices fueled such problems as cost forecasting. This research develops a time series model using the Box-Jenkins approach and material price time series data in Korea in order to forecast trends in the unit prices of required materials. Building information modeling (BIM) approaches are also used to analyze injection times of construction resources and to conduct quantity take-off so that total material prices can be forecast. To determine an optimal time series model for forecasting price trends, comparative analysis of predictability of tentative autoregressive integrated moving average (ARIMA) models is conducted. The proposed BIM-based time series forecasting model can help to deal with sudden changes in economic conditions by estimating material prices that correspond to resource injection times.

• Structural Engineering and Mechanics
• /
• 제21권6호
• /
• pp.621-634
• /
• 2005

A direct transfer substructure method is presented in this paper for analyzing the dynamic characteristics and the seismic random responses of a series reactor. This method combines the concept of FRF (frequency response function) and the transfer matrix algorithm with the substructure approach. The inner degrees of freedom of each substructure are eliminated in the process of reconstruction and the computation cost is reduced greatly. With the convenient solution procedure, the dynamic characteristics analysis of the structure is valid and efficient. Associated with the pseudo excitation algorithm, the direct transfer substructure method is applied to investigating the seismic random responses of the series reactor. The numerical results demonstrate that the presented method is efficient and practicable in engineering. Finally, a precise time integration method is employed in performing a time-history analysis on the series reactor under El Centro and Taft earthquake waves.

• 한국지능시스템학회논문지
• /
• 제21권6호
• /
• pp.730-736
• /
• 2011

본 연구는 순차적인 시계열 자료들에서 가장 최근의 추세가 반영될 수 있는 패턴분류 모델을 설계하였다. 의사결정을 지원하는 데이터마이닝 패턴분류 모델을 설계할 때 통계 기법과 인공지능 기법을 융합한 모델들이 기존의 모델보다 우수함을 입증하였다. 특히 퍼지이론과 융합된 패턴분류 모델들의 적중률이 상대적으로 더 향상되었다. 예를 들어, 통계적 이론을 기반으로 한 SVM모델과 퍼지소속함수와의 결합, 혹은 신경망과 FCM을 결합한 모델들의 성능이 우수하였다. 실험에서 사용한 패턴분류 모델들은 BPN, PNN, FNN, FCM, SVM, FSVM, Decision Tree, Time Series Analysis, Regression Analysis 등이다. 그리고 데이터베이스는 시계열 속성을 지닌 금융시장의 경제지표 DB(한국, KOSPI200 데이터베이스)와 병원 응급실의 부정맥환자에 대한 심전도 DB(미국 MIT-BIH 데이터베이스)들을 사용하였다.

• 산업경영시스템학회지
• /
• 제41권1호
• /
• pp.84-93
• /
• 2018

In this paper, we propose an Elman recurrent neural network to predict and analyze a time series of power energy consumption. To this end, we consider the volatility of the time series and apply the sample variance and the detrended fluctuation analyses to the volatilities. We demonstrate that there exists a correlation in the time series of the volatilities, which suggests that the power consumption time series contain a non-negligible amount of the non-linear correlation. Based on this finding, we adopt the Elman recurrent neural network as the model for the prediction of the power consumption. As the simplest form of the recurrent network, the Elman network is designed to learn sequential or time-varying pattern and could predict learned series of values. The Elman network has a layer of "context units" in addition to a standard feedforward network. By adjusting two parameters in the model and performing the cross validation, we demonstrated that the proposed model predicts the power consumption with the relative errors and the average errors in the range of 2%~5% and 3kWh~8kWh, respectively. To further confirm the experimental results, we performed two types of the cross validations designed for the time series data. We also support the validity of the model by analyzing the multi-step forecasting. We found that the prediction errors tend to be saturated although they increase as the prediction time step increases. The results of this study can be used to the energy management system in terms of the effective control of the cross usage of the electric and the gas energies.

• 한국융합학회논문지
• /
• 제12권10호
• /
• pp.9-17
• /
• 2021

최근 빅데이터 분석, 인공지능, 기계학습 등의 발전으로 인해서 데이터를 과학적으로 분석하는 기술이 발전하고 있으며 이는 의사결정 문제를 최적으로 해결해주고 있다. 그러나 특정 분야의 경우에는 데이터의 양이 부족해서 과학적 방식에 적용하는 것이 어렵다. 예컨대 부동산과 같은 데이터는 데이터 발표 시점이 최근이거나 비 유동성 자산이다 보니 발표 주기가 긴 경우가 많다. 따라서 본 연구에서는 이런 문제점을 극복하기 위해서 TimeGAN 모형을 통해 기존의 시계열의 확장 가능성에 대해서 연구하고자 한다. 이를 위해 부동산과 관련된 총 45개의 시계열을 데이터 셋에 맞게 2012년부터 2021년까지 주 단위로 데이터를 수집하고 시계열 간의 상관관계를 고려해서 총 15개의 최종 시계열을 선정한다. 15개의 시계열에 대해서 TimeGAN 모형을 통해 데이터 확장을한 결과, PCA 및 T-SNE 시각화 알고리즘을 통해 실제 데이터와 확장 데이터 간의 통계적 분포가 유사하다는 것을 확인할 수 있었다. 따라서 본 논문을 통해서 데이터의 과적합 또는 과소적합이라는 한계점을 극복할 수 있는 다양한 실험이 연구되기를 기대한다.

• 한국건설관리학회:학술대회논문집
• /
• 한국건설관리학회 2002년도 학술대회지
• /
• pp.64-71
• /
• 2002

This research developed construction cost forecasting model using Building Construction Cost Index, time series analysis and Artificial Neural Networks. By this model, we could calculate the forecasted values of construction cost precisely and efficiently. And we also could find out that the standard deviation of forecasted values is 0.375 and it is a very exact result, so the standard deviation is just 0.33 percent of 112.28, the average of Building Construction Cost Index. And it show more exact forecasting result in comparison with Time Series Analysis.