• 전자공학회논문지C
• /
• 제34C권6호
• /
• pp.71-78
• /
• 1997

In this paper, a new neural network training algorithm will be devised for function approximator with good generalization characteristics and tested with the time series prediction problem using santaFe competition data sets. To enhance the generalization ability a constraint term of hidden neuraon activations is added to the conventional output error, which gives the curvature smoothing characteristics to multi-layer neural networks. A hybrid learning algorithm of the error-back propagation and Hebbian learning algorithm with weight decay constraint will be naturally developed by the steepest decent algorithm minimizing the proposed cost function without much increase of computational requriements.

• Journal of the Korean Data and Information Science Society
• /
• 제10권2호
• /
• pp.329-340
• /
• 1999

We consider a Bayesian forcasting method for the analysis of repeated surveys. It is assumed that the parameters of the superpopulation model at each time follow a stochastic model. We propose Bayesian prediction procedures for the finite population total under multiprocess dynamic generalized linear models. The multiprocess dynamic model offers a powerful framework for the modelling and analysis of time series which are subject to a abrupt changes in pattern. Some numerical studies are provided to illustrate the behavior of the proposed predictors.

• International Journal of Contents
• /
• 제15권4호
• /
• pp.113-119
• /
• 2019

It has been reported that large amounts of information on agri-foods were delivered to consumers through television and social networks, and the information may influence consumers' behavior. The purpose of this paper was first to analyze relations of social network service and broadcasting program on paprika consumption in the aspect of amounts to purchase and identify potential factors that can promote paprika consumption; second, to develop prediction models of paprika consumption by using structured and unstructured big data. By using data 2010-2017, cross-correlation and time-series prediction algorithms (autoregressive exogenous model and vector error correction model), statistically significant correlations between paprika consumption and television programs/shows and blogs mentioning paprika and diet were identified with lagged times. When paprika and diet related data were added for prediction, these data improved the model predictability. This is the first report to predict paprika consumption by using structured and unstructured data.

• International journal of advanced smart convergence
• /
• 제7권4호
• /
• pp.75-83
• /
• 2018

Predictive IoT Sensor Algorithm is a technique of data science that helps computers learn from existing data to predict future behaviors, outcomes, and trends. This algorithm is a cloud predictive analytics service that makes it possible to quickly create and deploy predictive models as analytics solutions. Sensors and computers collect and analyze data. Using the time series prediction algorithm helps to predict future temperature. The application of this IoT in industrial environments like power plants and factories will allow organizations to process much larger data sets much faster and precisely. This rich source of sensor data can be networked, gathered and analyzed by super smart software which will help to detect problems, work more productively. Using predictive IoT technology - sensors and real-time monitoring - can help organizations exactly where and when equipment needs to be adjusted, replaced or how to act in a given situation.

• International Journal of Computer Science & Network Security
• /
• 제22권1호
• /
• pp.55-60
• /
• 2022

Churn prediction is a critical long-term problem for many business like music, games, magazines etc. The churn probability can be used to study many aspects of a business including proactive customer marketing, sales prediction, and churn-sensitive pricing models. It is quite challenging to design machine learning model to predict the customer churn accurately due to the large volume of the time-series data and the temporal issues of the data. In this paper, a parallel artificial neural network is proposed to create a highly-accurate customer churn model on a large customer dataset. The proposed model has achieved significant improvement in the accuracy of churn prediction. The scalability and effectiveness of the proposed algorithm is also studied.

• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2014년도 추계학술대회
• /
• pp.701-704
• /
• 2014

본 논문에서는 음료 판매 데이터 분석 및 판매량을 예측하는 방법을 제안하고자 하였다. 이를 위해 날씨와 음료 판매량이 상관관계가 있다고 가정하고, 온도, 습도를 입력으로 하여 판매량을 출력으로 하는 다항식 함수 관계를 모델링하였다. 본 논문에서는 제안한 방식의 유용성을 확인하기 위해 카페의 음료 판매 데이터를 2014년 2월부터 약 4개월 동안 수집하였고, 판매량 예측 알고리즘의 성능이 우수함을 확인하였다.

• 한국ITS학회 논문지
• /
• 제22권5호
• /
• pp.1-18
• /
• 2023

최근 빅데이터 및 딥러닝 기술의 발전으로 다양한 교통정보가 널리 수집 및 활용되고 있다. 특히 시계열 특성을 갖는 교통정보 예측 분야에서는 장단기 메모리(long short term memory, LSTM)가 널리 사용되고 있다. LSTM에 입력되는 시계열 데이터의 추세, 계절성, 주기 등이 상이하기 때문에 시계열 데이터를 기반으로 한 예측 모델에서도 데이터의 특성에 따라 하이퍼 파라미터의 적합한 값을 찾는 시행착오법이 필수적이다. 이에 적합한 하이퍼 파라미터를 찾는 방법론이 정립된다면, 정확도가 높은 모델 구성에 소요되는 시간을 줄일 수 있다. 따라서, 본 연구에서는 국내 고속도로 차량검지기 데이터와 LSTM을 기반으로 교통정보 예측 모델을 개발하였으며, LSTM의 하이퍼 파라미터별 평가지표 변화를 통해 예측 결과에 미치는 영향평가를 수행하였다. 또한, 이를 기반으로 교통분야에서 고속도로 교통정보 예측에 적합한 하이퍼 파라미터를 찾는 방법론을 제시하였다.

• 한국지하수토양환경학회지:지하수토양환경
• /
• 제19권3호
• /
• pp.123-133
• /
• 2014

It is important to predict the groundwater level fluctuation for effective management of groundwater monitoring system and groundwater resources. In the present study, three different time series models for the prediction of groundwater level in response to rainfall were built, those are transfer function noise model (TFNM), artificial neural network (ANN), and adaptive neuro fuzzy interference system (ANFIS). The models were applied to time series data of Boen, Cheolsan, and Hongcheon stations in National Groundwater Monitoring Network. The result shows that the model performance of ANN and ANFIS was higher than that of TFNM for the present case study. As lead time increased, prediction accuracy decreased with underestimation of peak values. The performance of the three models at Boen station was worst especially for TFNM, where the correlation between rainfall and groundwater data was lowest and the groundwater extraction is expected on account of agricultural activities. The sensitivity analysis for the input structure showed that ANFIS was most sensitive to input data combinations. It is expected that the time series model approach and results of the present study are meaningful and useful for the effective management of monitoring stations and groundwater resources.

• Journal of Positioning, Navigation, and Timing
• /
• 제13권1호
• /
• pp.111-115
• /
• 2024

This study builds a machine learning model optimized for clocks among various techniques in the field of artificial intelligence and applies it to clock stabilization or synchronization technology based on atomic clock noise characteristics. In addition, the possibility of providing stable source clock data is confirmed through the characteristics of machine learning predicted values during holdover of atomic clocks. The proposed machine learning model is evaluated by comparing its performance with the AutoRegressive Integrated Moving Average (ARIMA) model, an existing statistical clock prediction model. From the results of the analysis, the prediction model proposed in this study (MSE: 9.47476) has a lower MSE value than the ARIMA model (MSE: 221.2622), which means that it provides more accurate predictions. The prediction accuracy is based on understanding the complex nature of data that changes over time and how well the model reflects this. The application of a machine learning prediction model can be seen as a way to overcome the limitations of the statistical-based ARIMA model in time series prediction and achieve improved prediction performance.

• 산업경영시스템학회지
• /
• 제43권1호
• /
• pp.123-131
• /
• 2020

There have been a lot of studies in the past for the method of predicting the failure of a machine, and recently, a lot of researches and applications have been generated to diagnose the physical condition of the machine and the parts and to calculate the remaining life through various methods. Survival models are also used to predict plant failures based on past anomaly cycles. In particular, special machine that reflect the fluid flow and process characteristics of chemical plants are connected to hundreds or thousands of sensors, so there are not many factors that need to be considered, such as process and material data as well as application of derivative variables. In this paper, the data were preprocessed through time series anomaly detection based on unsupervised learning to predict the abnormalities of these special machine. Next, clustering results reflecting clustering-based data characteristics were applied to produce additional variables, and a learning data set was created based on the history of past facility abnormalities. Finally, the prediction methodology based on the supervised learning algorithm was applied, and the model update was confirmed to improve the accuracy of the prediction of facility failure. Through this, it is expected to improve the efficiency of facility operation by flexibly replacing the maintenance time and parts supply and demand by predicting abnormalities of machine and extracting key factors.