1,720,973 research outputs found
?????? ????????? ?????? ???????????? ????????? ?????? ?????? ????????? ?????? ?????? ???????????? ?????? ?????? ?????? ?????? ??????
Full text linkDepartment of Technology and Innovation ManagementMany machine learning applications are being employed to forecast weather conditions. In this paper, we focus more on small-scale weather forecasts with limited meteorological observation data. When oil refinery companies in non-oil-producing countries import crude oil by VLCCs (Very Large Crude Carriers), VLCCs unload crude oil to onshore storage tanks using SPM (Single Point Buoy Mooring System). Weather conditions in the offshore area where loading buoys are anchored are critical in determining whether unloading process is possible. The current practice of such decision making relies mostly on human experiences, and the predictive accuracy of the current practice is reported as about 75%. We tested machine learning methods to see if these methods can increase predictive accuracy in this problem of classification, the possibility of unloading given weather conditions such as wave heights, wind speeds, and wind directions. The results of our analysis showed that random forest and XGBoost have much better performance (more than 90%) than support vector machines and logistic regression in predicting unloading conditions in the time range from one hour to three days.clos
?????? ????????? ??????????????? ?????? ?????? ?????? ??? ??????
No full textGraduate School of Interdisciplinary Management (Business Analytics)PURPOSE: The purpose of this study is to investigate the various factors and to relatively analyze the importance according to the age group which affects sleep disorder by using machine learning techniques. From the initial 20 factors, a total of 14 factors are extracted through correlation analysis and variable selection method and then evaluated from 3 categories: Socio-demographic Factors, Health Behavior and Status, and Health-related Diseases. Based on this research, it aims to find comprehensive implications for the prevention and treatment of sleep disorder by age group.
METHODS: This study applied in 3,267 adults in the year of 2013 to 2014 by NHANES. The dependent variable which was used in this study is ???Sleep disorder???. First, rank correlation analysis (detecting multicollinearity) and stepwise selection method were performed to extract only meaningful variables and optimize the model. Next, contingency table and exploratory data analysis were performed to understand the frequency and basic information on sleep disorder. Finally, to relatively compare probabilities of sleep disorder according to age group, Logistic regression analysis was performed to determine the difference in probability after taking 3 datasets consisting of young people, middle-aged and elderly. And the importance of variables was verified by random forest.
RESULTS: ???Income level??? and ???Occupational status??? in socio-demographic factor, ???BMI index???, ???Depression experience??? and ???Excessive drinking', in health behavior and status, ???Hypertension???, ???High cholesterol??? and ???Diabetes??? in health-related disease had the strong influences on sleep disorder.
In middle-aged and elderly, high-income earners were about over 1.5 times likely to experience sleep disorder than low-income earners. For the unemployed, the probability of sleep disorder was 2.49 times higher for middle-aged and 1.41 times higher for elderly. BMI index has significantly affected by all ages except for young people. Especially, middle-aged with severe obese was greatly high, the probability of sleep disorder was up to 7 times. When people experienced depression, all ages were more than twice as likely to experience sleep disorder. Experience of depression is characterized by an absolute risk regardless of age. Almost diseases have a significant Influence on sleep disorder, but hypertension, high cholesterol, and diabetes were derived as representative diseases. Nonetheless, analysis showed that the significance was a little different for each age group.
CONCLUSION: This study is valuable as it provides useful information related to early diagnosis and social support about sleep disorder. When performing tasks and missions to prevent sleep disorder, its effectiveness can be maximized by selection and concentration factors according to each age group. Based on this study, I hope that it will contribute to the proper identification of the causative factors for the age group and further the establishment of preventive policies.ope
An Exploration of the Relationship between Boundary Spanning and Organizational Performance
No full textTechnology ManagementIn this paper, we present a study that examines how individuals who take the role of boundary spanner affect organizational innovativeness inside the team. Recently, there has been growing attention from burgeoning interests in open innovation and interdisciplinary R&D on boundary spanning and its impact on innovative culture or the capability of organizations. Boundary spanning is concerned with detecting internal or external information and then creating networks that connect between the environment and the organization. Such informational boundary spanners successfully translate acquired information and knowledge across communication boundaries. Therefore, they are considered key players of open innovation in many cases. To fill this role, they are usually aware of contextual conditions on both sides of the boundary and able to control the situation inside the firm.
For organizational innovativeness, we consider ambidexterity and absorptive capacity as theoretical foundations of our research. Ambidexterity refers to an organizational characteristic that pursues the balance between exploration of new knowledge and exploitation of existing knowledge. It is not counterintuitive that boundary spanning is associated with the activities of exploration as they are intended for tapping into diverse expertise and insights.
Our research model posits associations among vertical and horizontal boundary spanning within an organization, organizational combinative capabilities, and ambidexterity. We expect that this study can provide a better understanding on the dynamic mechanism of boundary spanning and the role of innovation leaders and also an insight into the questions: what is the bottleneck in the innovation process of our company? And how can we overcome the obstacles? Specifically, we will examine the relation among (1) Boundary spanning, (2) Diversity inside the unit, (3) Empowerment, (4) Ambidexterity, (5) Organizational performance. Thus, the main goal of this research is to examine whether the organizational performance varies as a results of boundary spanning roles which could be influenced by the diversity of the unit and empowerment.
The main method of our study was survey of professionals working in R&D departments. After reviewing relevant literature and selecting a pool of items, we conducted a survey. The questionnaire distributed randomly, and we mainly used survey instruments adopted from prior works. All components inside construct were measured with multi-item scales. Boundary spanners, ambidexterity, power relation, diversity, and performance were the latent variables. To remove the common method bias, we used Modern MTMM technique and Harman’s single-factor test. Also we examined differences between non-response biases. After checking the construct and content validity, and the reliability of the instruments, we employed PLS (partial least squares regression) analysis to find out the relations among variables.ope
Social Value Creation of Social Entrepreneurship: An Empirical Analysis
No full textDepartment Of Management EngineeringAlthough the importance of social entrepreneurship received growing attention from both scholars and practitioners as an alternative for the traditional public sector organizations, previous studies were too focused on the definition of social entrepreneurship and most of those studies were done as a case study. This paper hopes to fill the gap in the field of social entrepreneurship study by introducing alternative data source B-corporations and conduct an empirical study to test the result from previous studies on quantitative measure. By defining performance of social entrepreneurship as a social value creation, this study aimed to find out the factors that influence the social value creation. Result of this study implies that it is better for social entrepreneurship to focus on single social value rather than try to solve various social problems simultaneously and there are limitations on traditional accounting practice on measuring social entrepreneurships value creation.ope
Flipped Learning: An Empirical Study on the Inhibitors of Disruptive Innovation
No full textDepartment of Management EngineeringMany researchers have been highly expecting the change in the traditional education market to flipped learning. Although much previous research has investigated the benefits of flipped learning, expected disruption of teaching and learning practices has not yet come to fruition. Transforming from traditional educating systems to flipped learning mismatches the object of teaching and learning. Moreover, the diffusion of flipped learning is slow in progress and there are some underlying inhibitors of disruptive innovation. This paper aims to explain why flipped learning has not been speedily diffused in terms of disruptive innovation. We will empirically study the main factors ? path dependency, perceived efficiency, and perceived risk ? that might hinder the diffusion of flipped learning. Also, we will analyze these inhibitors through the survey conducted on students in the university setting. Our findings suggest that students who perceive risk of flipped learning would be path dependent on the traditional lecture, however, students would adopt flipped learning when they perceive its efficiency. Overall, our study would contribute to providing directions of the future education market.ope
???????????? ?????????????????? ??????????????? ?????? ?????? ??????
No full textSchool of Business Administration (Management Engineering)clos
Forecasting Korean LNG import price using ARIMAX, VECM, LSTM and hybrid models
Full text linkDepartment of Management Technology and Innovation ManagementIn this paper, an optimal forecasting model for the South Korean LNG import price was explored by combining an econometric model, a machine learning model, and a hybrid model. The autoregressive moving average model with extrinsic inputs model (ARIMAX) and VECM were the econometric models, and LSTM was the selected machine learning model. ARIMAX-LSTM and VECM-LSTM were used as hybrid models. Various independent variables, such as the Dubai oil price, European gas price, Australian Newcastle coal price, US natural gas price, Japanese liquified natural gas price and system marginal price in Korea were used for forecasting models. As it was proved that granger causality of each independent variables toward South Korean LNG import price is stronger in the order of the Dubai oil price, European gas price, Australian Newcastle coal price, US natural gas price, Japanese liquified natural gas price and SMP, the variables used for forecasting were added one by one in the order of strong granger causality. Optimal lags were derived from VECM analysis for each variable combination and these were used for VECM and LSTM prediction.
As a result of forecasting, 6 LSTM models, 4 VECM-LSTM were ranked in the top 10 forecasting models out of the total 90 models. Single econometric models were not included in the list. The best forecasting model was the LSTM with Dubai oil price, European gas price, Australian Newcastle coal price, US natural gas price, and Japanese liquified natural gas price with lag of 6, and its mean absolute percentage error (MAPE) was 3.5209. In addition, because LNG price forecasting is more important when price fluctuation is high, forecasting models were employed for 11 months with high fluctuation among the test periods. Seven hybrid models, one LSTM models, and two ARIMAX models were ranked in the top 10 forecasting models. VECM-LSTM using Dubai oil price with lag of 5 was derived as the best model with a MAPE of 4.9360. As a result of two forecasting analyses for both the whole and high fluctuation periods, we found that LSTM using Dubai oil price, European gas price, Australian Newcastle coal price, US natural gas price, and Japanese liquified natural gas price with a lag of 6 and VECM-LSTM using Dubai oil price, European gas price with a lag of 5 were ranked within the third best for both tests. Of the two models, the VECM-LSTM is in particular considered as the optimal model in that it has both high forecast accuracy and interpretability.ope
?????? ???????????? ????????? ?????? ??????????????? ?????? : ????????? ?????? ?????? ??????????????? ????????????
Full text linkDepartment of Management EngineeringFirms participating in printer industries have invested their constrained resources into technology development in order to sustain their competitiveness in the industry. Considering the fast-changing market circumstances, each firm???s own investment decisions on technology portfolio may directly affect their performance.
In this study, we analyzed patent data, namely number of forward citations and technological classification data (CPC). Using this data, the technological portfolio of a specific firm can be identified, which can further help our understanding on firms??? R&D investment strategies. Number of studies mainly focused on patent class combinations of individual technology level, but portfolios of patent class at a firm level have been understudied.
In this study, we tracked the change of class composition within each firms??? technological patents??? portfolio and attempted to identify practical and theoretical implications to portfolio management. We utilized Entropy Index, Co-occurrence and cosine similarities measurements for each indicating diversification, patent scope and portfolio similarities within each patents??? classification subclasses. Additionally, performance evaluation of each portfolio is conducted using forward citation data.
This paper shows that in-depth patent data analysis can allow us to explore deeper insights at various levels, individual technology, products and product lines, and firms sufficing different stories.ope
Technological Innovation Performance Analysis Using Multilayer Networks: Evidence from the Printer Industry
Full text linkDepartment of Management EngineeringThe importance of collaboration and technology boundary spanning has been emphasized in other inquiries into technological innovation. Therefore, this research project first tried to investigate the effect of collaboration on technology boundary spanning. Then, we investigated the effect of collaboration and technology boundary spanning on technological innovation within a firm by using a multilayer network to analyze patent data. The aim of this paper is to provide new insight into the process of analyzing patent data using multilayer networks. This empirical study is based on a sample of 408 firms within the printer industry from 1996 to 2005.
Starting with a theoretical discussion of R&D collaboration, technology boundary spanning and innovation performance, the importance of a firm???s collaboration and technology boundary spanning in its technology innovation performance was empirically analyzed using patent data. We followed changes in collaboration networks, technology class networks and the connection between them and tried to find the meaning of those changes in firms??? technology innovation performances. We used degree centrality within the collaboration network and the ratio of collaborated patents to the total number of patents in order to measure a firm???s collaboration and formulated technology boundary spanning represented by exploitation and exploration by using edges of the multilayer network. As dependent variables, we used the number of patents and the average number of citations received over three, five, and 10 years to measure the firm???s quantitative and qualitative innovation performance respectively.
The results of the analysis can be summarized as follows: a firm???s collaboration has positive effects on both exploitation and exploration. Firms with more collaborations show higher quantitative innovation performances while firms with more collaborations exhibit lower qualitative innovation performance. Exploitation has a positive impact on a firm???s quantitative innovation performance while exploration has negative effects on a firm???s quantitative innovation performance. The relationship between a firm???s exploration activities and a firm???s qualitative innovation performance manifests as an inverted U-shape. On the other hand, a firm???s exploitation activities have a U-shape relationship with the firm???s qualitative innovation performance.
The implication of this study is that multilayer networks can be used to analyze patent data. This study used multilayer networks to formulate the exploitation and exploration only. However, in further research it can be utilized to find the hub firms that fuse technologies.clos
ENERGY CONSUMPTION FORECASTING IN SOUTH KOREA USING MACHINE LEARNING ALGORITHMS
No full textGraduate School of Interdisciplinary Management (Business Analytics)In the changing environmental climate, accurate prediction of energy consumption is crucial in meeting the energy demand. Traditionally, classic econometric and statistical models are used to forecast energy consumption. These models may have limitations in an increasingly fast-changing energy market, which require big-data analysis of energy consumption patterns and relevant variables using complex mathematical tools. In current literature, there are minimal comparison studies reviewing machine-learning algorithms to predict energy consumption in South Korea. To bridge this gap, this thesis compared three different machine-learning algorithms, namely the Random Forest (RF) model, XGBoost model and Long Short-Term Memory (LSTM) model. These algorithms were applied in Period 1 (prior to the onset of the COVID-19 pandemic) and Period 2 (after the onset of COVID-19 pandemic). Period 1 was characterized by an upward trend of energy consumption, while Period 2 showed a reduction in energy consumption. LSTM performed best in its prediction power specifically in Period 1, and RF outperformed the other models in Period 2. Findings therefore suggested that selective use of algorithms based on the energy consumption patterns will likely yield the most predictive power. This confirms the value of machine-learning algorithms when used in the appropriate setting and can provide reliable information to stakeholders involved in the planning and implementation of energy solutions.clos
- …
