• 산업경영시스템학회지
• /
• 제45권2호
• /
• pp.20-29
• /
• 2022

This study intends to present a traffic node-based and link-based accident prediction models using XGBoost which is very excellent in performance among machine learning models, and to develop those models with sustainability and scalability. Also, we intend to present those models which predict the number of annual traffic accidents based on road types, weather conditions, and traffic information using XGBoost. To this end, data sets were constructed by collecting and preprocessing traffic accident information, road information, weather information, and traffic information. The SHAP method was used to identify the variables affecting the number of traffic accidents. The five main variables of the traffic node-based accident prediction model were snow cover, precipitation, the number of entering lanes and connected links, and slow speed. Otherwise, those of the traffic link-based accident prediction model were snow cover, precipitation, the number of lanes, road length, and slow speed. As the evaluation results of those models, the RMSE values of those models were each 0.2035 and 0.2107. In this study, only data from Sejong City were used to our models, but ours can be applied to all regions where traffic nodes and links are constructed. Therefore, our prediction models can be extended to a wider range.

• Journal of Forest and Environmental Science
• /
• 제40권1호
• /
• pp.15-23
• /
• 2024

This research aimed to assess the possibility of detecting forest degradation using time-series satellite imagery and three different deep learning-based change detection techniques. The dataset used for the deep learning models was composed of two sets, one based on surface reflectance (SR) spectral information from satellite imagery, combined with Texture Information (GLCM; Gray-Level Co-occurrence Matrix) and terrain information. The deep learning models employed for land cover change detection included image differencing using the Unet semantic segmentation model, multi-encoder Unet model, and multi-encoder Unet++ model. The study found that there was no significant difference in accuracy between the deep learning models for forest degradation detection. Both training and validation accuracies were approx-imately 89% and 92%, respectively. Among the three deep learning models, the multi-encoder Unet model showed the most efficient analysis time and comparable accuracy. Moreover, models that incorporated both texture and gradient information in addition to spectral information were found to have a higher classification accuracy compared to models that used only spectral information. Overall, the accuracy of forest degradation extraction was outstanding, achieving 98%.

• Journal of Information Processing Systems
• /
• 제16권4호
• /
• pp.809-819
• /
• 2020

Gender prediction accuracy increases as convolutional neural network (CNN) architecture evolves. This paper compares voting and ensemble schemes to utilize the already trained five CNN models to further improve gender prediction accuracy. The majority voting usually requires odd-numbered models while the proposed softmax-based voting can utilize any number of models to improve accuracy. The ensemble of CNN models combined with one more fully-connected layer requires further tuning or training of the models combined. With experiments, it is observed that the voting or ensemble of CNN models leads to further improvement of gender prediction accuracy and that especially softmax-based voters always show better gender prediction accuracy than majority voters. Also, compared with softmax-based voters, ensemble models show a slightly better or similar accuracy with added training of the combined CNN models. Softmax-based voting can be a fast and efficient way to get better accuracy without further training since the selection of the top accuracy models among available CNN pre-trained models usually leads to similar accuracy to that of the corresponding ensemble models.

• Journal of Biosystems Engineering
• /
• 제29권1호
• /
• pp.65-70
• /
• 2004

This study was carried out to verify and predict the soil informations such as the contents of organic matter(OM) and Mg and pH of the soil. The predictability of spacial variation in the paddy field was examined by analyzing the various soil information. The prediction models for the OM pH, and Mg, were developed using inverse distance weighted (IDW), triangulated irregular network(TIN) and Kriging model. The determination of coefficients of linear and spherical Kriging models were 0.756 and 0.578, respectively, and were very low in comparison with other soil information. For IDW and TIN model, the determination of coefficients were 1.000 and hence the performance of the models was found to be excellent. The developed models were validated using unknown soil sample obtained In 2000 and 2001. From the analysis of relationship between the measured pH and predicted 0.9353. For prediction of Mg, the determination of coefficient is more than 0.8. Since the determination of coefficients of developed models for OM were relatively low, it may be difficult to predict the content of OM using the developed models. For further study, the additional works to enhance the performance of the prediction models for soil information are required.

• 한국문헌정보학회지
• /
• 제49권3호
• /
• pp.273-289
• /
• 2015

This study examined changes in citation index scores and rankings of thirty-five chemistry faculty members at Seoul National University using different co-authorship credit allocation models. Using 1,436 Web of Science papers published between 2007 and 2013, we applied the inflated, fractional, harmonic, network-based allocation, and harmonic+ models to calculate faculty's h-, R-, and normalization of h- and R- index scores and rankings. The harmonic+ model, which is based on our belief that contribution of primary authors should be the same regardless of collaboration, is designed to minimize the penalty for research collaboration imposed by harmonic and NBA models by boosting the contribution of collaborating primary authors to be on the equal footing with single authors. Although citation rankings by different models are correlated with each other within the same type of citation indicator, rankings of many faculty members changed across models, suggesting the importance of an accurate and relevant authorship credit allocation model in the citation assessment of researchers. The study also found that authorship patterns in conjunction with citation counts are important factors for robust authorship models such as harmonic and NBA, and harmonic+ model may be beneficial for collaborating primary authors. Future research that reexamines the models with updated empirical data would provide further insights into the robustness of the models.

• 대한원격탐사학회:학술대회논문집
• /
• 대한원격탐사학회 2004년도 Proceedings of ISRS 2004
• /
• pp.622-625
• /
• 2004

This paper presents multi-source spatial data integration models based on probability theory for landslide hazard assessment. Four probabilistic models such as empirical likelihood ratio estimation, logistic regression, generalized additive and predictive discriminant models are proposed and applied. The models proposed here are theoretically based on statistical relationships between landslide occurrences and input spatial data sets. Those models especially have the advantage of direct use of continuous data without any information loss. A case study from the Gangneung area, Korea was carried out to quantitatively assess those four models and to discuss operational issues.

• Journal of the Korean Data and Information Science Society
• /
• 제14권4호
• /
• pp.825-836
• /
• 2003

Based on customer information and financing processes in capital market, we derived individual models by applying multi-layered perceptrons, MDA, and decision tree. Further, the results from the existing single models were compared with the results from the integrated model that was developed using genetic algorithm. This study contributes not only to verifying the existing individual models and but also to overcoming the limitations of the existing approaches. We have depended upon the approaches that compare individual models and search for the best-fit model. However, this study presents a methodology to build an integrated data mining model using genetic algorithm.

• Asia pacific journal of information systems
• /
• 제12권2호
• /
• pp.69-98
• /
• 2002

With the development of the Internet, electronic commerce, electronic markets, and digital economy, new business paradigm and new ways of business have been emerging and developing. The development of right and robust business models for electronic markets is a key for e-business success. This paper reviews previous studies and successful cases for e-business models. This paper presents strategic factors such as the business value and the source of revenue, products and services, business processes and technologies, and the characteristics of markets and relationship with customers and partners as a framework for developing sustainable and robust business models.

• Asia pacific journal of information systems
• /
• 제26권1호
• /
• pp.66-79
• /
• 2016

The availability of detailed data on customers' online behaviors and advances in big data analysis techniques enable us to predict consumer behaviors. In the past, researchers have built purchase prediction models by analyzing clickstream data; however, these clickstream-based prediction models have had several limitations. In this study, we propose a new method for purchase prediction that combines information theory with machine learning techniques. Clickstreams from 5,000 panel members and data on their purchases of electronics, fashion, and cosmetics products were analyzed. Clickstreams were summarized using the 'entropy' concept from information theory, while 'random forests' method was applied to build prediction models. The results show that prediction accuracy of this new method ranges from 0.56 to 0.83, which is a significant improvement over values for clickstream-based prediction models presented in the past. The results indicate further that consumers' information search behaviors differ significantly across product categories.

• Journal of the Korean Data and Information Science Society
• /
• 제20권5호
• /
• pp.895-903
• /
• 2009

The penalized partial likelihood based on restricted maximum likelihood method has been widely used for the inference of frailty models. However, the standard-error estimate for frailty parameter estimator can be downwardly biased. In this paper we show that such underestimation can be corrected by using hierarchical likelihood. In particular, the hierarchical likelihood gives a statistically efficient procedure for various random-effect models including frailty models. The proposed method is illustrated via a numerical example and simulation study. The simulation results demonstrate that the corrected standard-error estimate largely improves such bias.