• Journal of Information Technology Applications and Management
• /
• v.31 no.1
• /
• pp.11-25
• /
• 2024

The purpose of this study is to investigate the improvement of prediction performance and changes in variable importance in SEM-ANN two-stage analysis. 366 cosmetics repurchase-related survey data were analyzed and the results were presented. The results of this study are summarized as follows. First, in SEM-ANN two-stage analysis, SEM and ANN models were trained with train data and predicted with test data, respectively, and the R² was showed. As a result, the prediction performance was doubled from SEM 0.3364 to ANN 0.6836. Looking at this degree of R² improvement as the effect size f² of Cohen (1988), it corresponds to a very large effect at 110%. Second, as a result of comparing changes in normalized variable importance through SEM-ANN two-stage analysis, variables with high importance in SEM were also found to have high importance in ANN, but variables with little or no importance in SEM became important in ANN. This study is meaningful in that it increased the validity of the comparison by using the same learning and evaluation method in the SEM-ANN two-stage analysis. This study is meaningful in that it compared the degree of improvement in prediction performance and the change in variable importance through SEM-ANN two-stage analysis.

• International Journal of Contents
• /
• v.5 no.1
• /
• pp.37-45
• /
• 2009

This paper empirically tests pecking order theory. Korean listed firms are used as the samples. On the whole we find supportive results for pecking order theory. The fixed effect model on the whole period shows that as pecking order theory suggests that debt ratio decreases as cash flow. ROA, physical assets, and firm size increase. Again, it is shown that corporate debt ratio significantly decreases as cash flow or ROA increases in every sub-sample, which coincides with the prediction of pecking order theory. Corporate debt ratio significantly decreases as physical assets or jinn size increases in case of the whole sample, pre-financial crisis period, and the sub-samples by q-ratio, which also supports the prediction of pecking order theory. Statistical significance of the coefficients of physical assets or firm size completely disappears after Korean financial crisis. Perhaps it is because the role of physical assets or firm size as a mitigator of information asymmetry significantly weakens after the financial crisis as Korean financial market becomes more transparent. For small firms only size variable is negatively and significantly related with debt to assets. It seems that size is an important factor for smaller firms in making financing decision.

• Journal of Welding and Joining
• /
• v.28 no.3
• /
• pp.104-113
• /
• 2010

In this study, to predict the microstructure in weld HAZ of low alloyed steel, prediction model for the phase transformation considering the influence of prior austenite grain size and cooling rate was developed. For this study, six low alloyed steels were designed and the effect of alloying elements was also investigated. In order to develop the prediction model for ferrite transformation, isothermal ferrite transformation behaviors were analyzed by dilatometer system and 'Avrami equation' which was modified to consider the effect of prior austenite grain size. After that, model for ferrite phase transformation during continuous cooling was proposed based on the isothermal ferrite transformation model through applying the 'Additivity rule'. Also, start temperatures of ferrite transformation were predicted by $A_{r3}$ considering the cooling rate. CCT diagram was calculated through this model, these results were in good agreement with the experimental results. After ferrite transformation, bainite transformation was predicted using Esaka model which corresponded most closely to the experimental results among various models. The start temperatures of bainite transformation were determined using K. J. Lee model. Phase fraction of martensite was obtained according to phase fractions of ferrite and bainite.

• Journal of KIISE:Computer Systems and Theory
• /
• v.27 no.8
• /
• pp.719-729
• /
• 2000

Prefetching aims at reducing memory latency by fetching, in advance, data that are likely to be requested by the processor in a near future. The effectiveness of prefetching is determined by how accurate the prediction on the needed instructions and data is. Most previous studies on prefetching were limited to proposing a particular prefetch scheme and its performance evaluation, paying little attention to theoretical aspects of prefetching. This paper focuses on the theoretical aspects of instruction prefetching. For this purpose, we propose a clairvoyant prefetch model that makes use of perfect history information. Based on this theoretical model, we analyzed upper limits on the prefetch prediction accuracies of the SPEC benchmarks. The results show that the prefetch prediction accuracy is very high when there is no cache. However, as the size of the instruction cache increases, the prefetch prediction accuracy drops drastically. For example, in the case of the spice benchmark, the prefetch prediction accuracy drops from 53% to 39% when the cache size increases from 2Kbyte to 16Kbyte (assuming 16byte block size). These results indicate that as the cache size increases, most localities are captured by the cache and that instruction prefetching based on the information extracted from the references that missed in the cache suffers from prediction inaccuracies

• Journal of Korea Multimedia Society
• /
• v.11 no.7
• /
• pp.956-965
• /
• 2008

H.264/AVC is the newest video coding standard of ITU-T VCEG and the ISO/IEC MPEG, offering a significant performance improvement over previous video coding standards. However, the computational complexity of H.264/AVC is drastically increased because of new technologies such as intra prediction, variable block size, quarter-pels motion estimation/compensation, etc. In this paper, we propose a fast intra prediction scheme which has two step processing. The first step is a fast block size decision which can be calculated only in one block without considering all cases of $4{\times}4$ block and $16{\times}16$ block. The complexity of the intra prediction can be reduced by using boundary difference values of macroblock. After selecting the block size, we can make mode decision using the neighbouring reference pixels and representative pixels of the block in the second step. The experimental results show that the proposed algorithm saved on the average 41.5% encoding time without any significant PSNR losses.

• The Korean Journal of Applied Statistics
• /
• v.34 no.5
• /
• pp.697-710
• /
• 2021

Relative error prediction is preferred over ordinary prediction methods when relative/percentile errors are regarded as important, especially in econometrics, software engineering and government official statistics. The relative error prediction techniques have been developed in linear/nonlinear regression, nonparametric regression using kernel regression smoother, and stationary time series models. However, random effect models have not been used in relative error prediction. The purpose of this article is to extend relative error prediction to some of generalized linear mixed model (GLMM) with panel data, which is the random effect models based on gamma, lognormal, or inverse gaussian distribution. For better understanding, the real auto insurance data is used to predict the claim size, and the best predictor and the best relative error predictor are comparatively illustrated.

• Journal of Information Processing Systems
• /
• v.17 no.2
• /
• pp.213-226
• /
• 2021

Given that most of the link prediction algorithms for signed social networks can only complete sign prediction, a novel algorithm is proposed aiming to achieve both link prediction and sign prediction in signed networks. Based on the structural balance theory, the local link tightness and global link tightness are defined respectively by using the structural information of paths with the step size of 2 and 3 between the two nodes. Then the total similarity of the node pair can be obtained by combining them. Its absolute value measures the possibility of the two nodes to establish a link, and its sign is the sign prediction result of the predicted link. The effectiveness and correctness of the proposed algorithm are verified on six typical datasets. Comparison and analysis are also carried out with the classical prediction algorithms in signed networks such as CN-Predict, ICN-Predict, and PSNBS (prediction in signed networks based on balance and similarity) using the evaluation indexes like area under the curve (AUC), Precision, improved AUC', improved Accuracy', and so on. Results show that the proposed algorithm achieves good performance in both link prediction and sign prediction, and its accuracy is higher than other algorithms. Moreover, it can achieve a good balance between prediction accuracy and computational complexity.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.8 no.1
• /
• pp.31-36
• /
• 2008

This paper presents adaptive learning data of evolvable neural networks (ENNs) for time series prediction of nonlinear dynamic systems. ENNs are a special class of neural networks that adopt the concept of biological evolution as a mechanism of adaptation or learning. ENNs can adapt to an environment as well as changes in the enviromuent. ENNs used in this paper are L-system and DNA coding based ENNs. The ENNs adopt the evolution of simultaneous network architecture and weights using indirect encoding. In general just previous data are used for training the predictor that predicts future data. However the characteristics of data and appropriate size of learning data are usually unknown. Therefore we propose adaptive change of learning data size to predict the future data effectively. In order to verify the effectiveness of our scheme, we apply it to chaotic time series predictions of Mackey-Glass data.

• Journal of Korean Institute of Industrial Engineers
• /
• v.28 no.2
• /
• pp.171-177
• /
• 2002

In this paper, we propose a new data fusion method to improve the performance of individual prediction models for time series data. Individual models used are ARIMA and neural network and their results are combined based on the weight reflecting the inverse of EWMA of squared prediction error of each individual model. Monte Carlo simulation is used to identify the situation where the proposed approach can take a vintage point over typical fusion methods which utilize MSE for weight. Study results indicate the following: EWMA performs better than MSE fusion when the data size is large with a relatively big amplitude, which is often observed in intra-cranial pressure data. Additionally, EWMA turns out to be a best choice among MSE fusion and the two individual prediction models when the data size is large with relatively small random noises, often appearing in tax revenue data.

• 제어로봇시스템학회:학술대회논문집
• /
• 2005.06a
• /
• pp.920-924
• /
• 2005

This paper presents adaptive learning data of evolvable neural networks (ENNs) for time series prediction of nonlinear dynamic systems. ENNs are a special class of neural networks that adopt the concept of biological evolution as a mechanism of adaptation or learning. ENNs can adapt to an environment as well as changes in the environment. ENNs used in this paper are L-system and DNA coding based ENNs. The ENNs adopt the evolution of simultaneous network architecture and weights using indirect encoding. In general just previous data are used for training the predictor that predicts future data. However the characteristics of data and appropriate size of learning data are usually unknown. Therefore we propose adaptive change of learning data size to predict the future data effectively. In order to verify the effectiveness of our scheme, we apply it to chaotic time series predictions of Mackey-Glass data.