• 한국정보시스템학회지:정보시스템연구
• /
• 제29권1호
• /
• pp.241-265
• /
• 2020

Purpose Various machine learning techniques are used to implement for predicting corporate credit. However, previous research doesn't utilize time series input features and has a limited prediction timing. Furthermore, in the case of corporate bond credit rating forecast, corporate sample is limited because only large companies are selected for corporate bond credit rating. To address limitations of prior research, this study attempts to implement a predictive model with more sample companies, which can adjust the forecasting point at the present time by using the credit score information and corporate information in time series. Design/methodology/approach To implement this forecasting model, this study uses the sample of 2,191 companies with KIS credit scores for 18 years from 2000 to 2017. For improving the performance of the predictive model, various financial and non-financial features are applied as input variables in a time series through a sliding window technique. In addition, this research also tests various machine learning techniques that were traditionally used to increase the validity of analysis results, and the deep learning technique that is being actively researched of late. Findings RNN-based stateful LSTM model shows good performance in credit rating prediction. By extending the forecasting time point, we find how the performance of the predictive model changes over time and evaluate the feature groups in the short and long terms. In comparison with other studies, the results of 5 classification prediction through label reclassification show good performance relatively. In addition, about 90% accuracy is found in the bad credit forecasts.

• 한국멀티미디어학회논문지
• /
• 제25권1호
• /
• 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.

• Journal of the Korean Data and Information Science Society
• /
• 제10권1호
• /
• pp.147-154
• /
• 1999

Using a noninformative prior and an inverted gamma prior, the Bayesian predictive density and the prediction intervals for a future observation or the p - th order statistic of n' future observations from the censord Pareto model have been obtained. In additions, numerical examples are given in order to illustrate the proposed predictive procedure.

• Geomechanics and Engineering
• /
• 제18권4호
• /
• pp.373-389
• /
• 2019

It is important to study the problem of durability for tunnel structures. As a main influence on the durability of tunnel structures, carbonation-induced corrosion is studied. For the complicated environment of tunnel structures, based on the data samples from real engineering examples, the intelligent method (genetic programming) is used to construct the service life prediction model of tunnel structures. Based on the model, the prediction of service life for tunnel structures in carbonation environments is studied. Using the data samples from some tunnel engineering examples in China under carbonation environment, the proposed method is verified. In addition, the performance of the proposed prediction model is compared with that of the artificial neural network method. Finally, the effect of two main controlling parameters, the population size and sample size, on the performance of the prediction model by genetic programming is analyzed in detail.

• East Asian Economic Review
• /
• 제27권3호
• /
• pp.213-242
• /
• 2023

This paper investigates whether the price of cryptocurrency is determined by the US dollar index, the price of investment assets such gold and oil, and the implied volatility of the KOSPI. Overall, the returns on cryptocurrencies are best predicted by the trading volume of the cryptocurrency both in-sample and out-of-sample. The estimates of gold and the dollar index are negative in the return prediction, though they are not significant. The dollar index, gold, and the cryptocurrencies seem to share characteristics which hedging instruments have in common. When investors take notice of the imminent market risks, they increase the demand for one of these assets and thereby increase the returns on the asset. The most notable result in the out-of-sample predictability is the predictability of the returns on value-weighted portfolio by gold. The empirical results show that the restricted model fails to encompass the unrestricted model. Therefore, the unrestricted model is significant in improving out-of-sample predictability of the portfolio returns using gold. From the empirical analyses, we can conclude that in-sample predictability cannot guarantee out-of-sample predictability and vice versa. This may shed light on the disparate results between in-sample and out-of-sample predictability in a large body of previous literature.

• 한국음향학회지
• /
• 제11권1호
• /
• pp.60-66
• /
• 1992

음성표본간의 자기상관관계를 살펴보면 인접한 양쪽 표본의 자기상관관계가 시간지연된 여러 차수의 것보다는 크게 나타난다. 그러므로 현재표본을 기준으로 몇 차 지연된 과거값을 이용하여 현재값을 에측하는 것보다는 바로 인접한 과거 및 미래의 표본에 의해 현재의 표본값을 예측하는 것이 더 효과적이다. 따라서 본 논문에서는 한 표본 미래값과 한 표본 과거값을 취하여 현재의 표본을 예측하는 파형부호화용 예측기를 새로이 제안하였다. 제안된 예측기는 CCITT-표준 ADPCM보다 예측이득이 최고 9dB까지 높게 얻어졌다.

• Journal of the Korean Data and Information Science Society
• /
• 제11권2호
• /
• pp.173-179
• /
• 2000

Interval prediction based on the empirical distribution function for the class of time series with time varying coefficients is discussed. To this end, strong mixing property of the model is shown and results due to Fotopoulos et. al.(1994) are employed. A simulation study is presented to assess the accuracy of the proposed interval predictor.

• Wind and Structures
• /
• 제38권2호
• /
• pp.147-160
• /
• 2024

The aerodynamic force is a significant component that influences the stability and safety of structures. It has unstable properties and depends on computer precision, making its long-term prediction challenging. Accurately estimating the aerodynamic traits of structures is critical for structural design and vibration control. This paper establishes an unsteady aerodynamic time series prediction model using Long Short-Term Memory (LSTM) network. The unsteady aerodynamic force under varied Reynolds number and angles of attack is predicted by the LSTM model. The input of the model is the aerodynamic coefficients of the 1 to n sample points and output is the aerodynamic coefficients of the n+1 sample point. The model is predicted by interpolation and extrapolation utilizing Unsteady Reynolds-average Navier-Stokes (URANS) simulation data of flow around a circular cylinder, square cylinder and airfoil. The results illustrate that the trajectories of the LSTM prediction results and URANS outcomes are largely consistent with time. The mean relative error between the forecast results and the original results is less than 6%. Therefore, our technique has a prospective application in unsteady aerodynamic force prediction of structures and can give technical assistance for engineering applications.

• 한국정보통신학회논문지
• /
• 제14권2호
• /
• pp.534-539
• /
• 2010

게놈 연구에서 수천 개의 특징들은 비교적 작은 샘플들로부터 모아진다. 게놈 연구의 목적은 미래 관찰들의 결과를 예측하는 분류기를 만드는 것이다. 분류기를 만들기 위해서는 특징 선택, 모델 선택 그리고 예측 평가 등의 3단계 과정을 거친다. 본 논문은 예측 평가에 초점을 맞추고 모든 슬라이드의 사분위수를 똑같게 맞추는 quantilenormalization 적용하여 마이크로어레이 데이터를 표준화 한 후 특징 선택에 앞서 예측 모델의 '진짜' 예측 에러를 평가하기 위해 몇 개의 방법들을 비교하는 시스템을 고안하고 방법들의 예측 에러를 비교 분석 하였다. LOOCV는 전체적으로 작은 MSE와 bias를 나타내었고, 크기가 작은 샘플에서 split 방법과 2-fold CV는 매우 좋지 않는 결과를 보였다. 계산적으로 번거로운 분석에 대해서는 10-fold CV가 LOOCV보다 오히려 더 낳은 경향을 보였다.

• Journal of Digestive Cancer Research
• /
• 제11권3호
• /
• pp.157-164
• /
• 2023

Prediction is a significant topic in clinical research. The development and validation of a prediction model has been increasingly published in clinical research. In this review, we investigated analytical methods and validation schemes for a clinical prediction model used in digestive cancer research. Deep learning and logistic regression, with split-sample validation as an internal or external validation, were the most commonly used methods. Furthermore, we briefly introduced and summarized the advantages and disadvantages of each method. Finally, we discussed several points to consider when conducting prediction model studies.