• Korea Trade Review
• /
• v.46 no.1
• /
• pp.169-187
• /
• 2021

This paper analyzes the feasibility of using machine learning and deep learning methods to forecast the income and employment using the strategic industries as well as investment, export, and exchange rates. The decision tree, artificial neural network, support vector machine, and deep learning models were used to forecast the income and employment in Busan. The following were the main findings of the comparison of their predictive abilities. First, the decision tree models predict the income and employment well. The forecasting values for the income and employment appeared somewhat differently according to the depth of decision trees and several conditions of strategic industries as well as investment, export, and exchange rates. Second, since the artificial neural network models show that the coefficients are somewhat low and RMSE are somewhat high, these models are not good forecasting the income and employment. Third, the support vector machine models show the high predictive power with the high coefficients of determination and low RMSE. Fourth, the deep neural network models show the higher predictive power with appropriate epochs and batch sizes. Thus, since the machine learning and deep learning models can predict the employment well, we need to adopt the machine learning and deep learning models to forecast the income and employment.

• Structural Engineering and Mechanics
• /
• v.89 no.3
• /
• pp.283-300
• /
• 2024

Machine learning (ML) models based on artificial neural network (ANN) and decision tree (DT) were developed for estimation of axial capacity of concrete columns reinforced with fiber reinforced polymer (FRP) bars. Between the design codes, the Canadian code provides better formulation compared to the Australian or American code. For empirical models based on elastic modulus of FRP, Hadhood et al. (2017) model performed best. Whereas for empirical models based on tensile strength of FRP, as well as all empirical models, Raza et al. (2021) was adjudged superior. However, compared to the empirical models, all ML models exhibited superior performance according to all five performance metrics considered. The performance of ANN and DT models were comparable in general. Under the present setup, inclusion of the transverse reinforcement information did not improve the accuracy of estimation with either ANN or DT. With selective use of inputs, and a much simpler ANN architecture (4-3-1) compared to that reported in literature (Raza et al. 2020: 6-11-11-1), marginal improvement in correlation could be achieved. The metrics for the best model from the study was a correlation of 0.94, absolute errors between 420 kN to 530 kN, and the range being 0.39 to 0.51 for relative errors. Though much superior performance could be obtained using ANN/DT models over empirical models, further work towards improving accuracy of the estimation is indicated before design of FRP reinforced concrete columns using ML may be considered for design codes.

• Journal of Korean Society of Transportation
• /
• v.17 no.5
• /
• pp.99-110
• /
• 1999

This study deals with the adaptability questions of O-D table estimation models. Its objectives are two-fold; (1) to estimate the characteristics of various O-D table estimation models(i.e. linear regression models. entropy models and statistic models) and (2) to find the model which estimates the O-D table with the best accuracy under the various data conditions. In Pursuing the above, this study gives the particular attentions to the test of the models, using the Sioux Falls network and equilibrium assignment method of MINUTP. The major findings are the followings. Firstly. it finds that the statistic models have the most goodness of fat among all models, if the required data are all Prepared. But it Presents that statistic models are the most sensitive against the underspecification and inconsistency problems of link data. Secondly, It shows that the linear regression models have the worst goodness of fat among all models. But the linear regression models are the most insensitive to the underspecification and inconsistency problems. Thirdly, THE/1 model of entropy model is sensitive against the underspecification and incon-sistency problems, but THE/2 model is insensitive. Finally, other informations like total volume, zonal Production and attraction volumes in 0-D table, help models to gain the better goodness of fit. Especially, in the statistic models. both the zonal production and attraction volume data are helpful to estimate the link volumes. It can be expected that the results dive some implications not only to the selection of optimal model under the various given data, but also to the development or modification of model.

• International Journal of Aeronautical and Space Sciences
• /
• v.16 no.2
• /
• pp.206-222
• /
• 2015

This work uses different finite element approaches to the free vibration analysis of reinforced shell structures, and a simplified model of a typical launcher with two boosters is used as an example. The results obtained using a refined one-dimensional (1D) beam model are compared to those obtained with commercial finite element software. The 1D models that are used in the present work are based on the Carrera Unified Formulation (CUF), which assumes a variable kinematic displacement field over the cross-sections of the beam. Two different sets of polynomials that correspond to Taylor (TE) or Lagrange (LE) expansions were used. The analyses focused on three reinforced structures: a stiffened panel, a reinforced cylinder and the complete structure of the launcher. The frequencies and natural modes obtained using one-dimensional models are compared to those obtained from classical finite element analysis. The classical FE models were built using a beam-shell or solid elements, and the results indicate that the refined beam models can in fact be used to investigate the behavior of very complex reinforced structures. These models can predict the shell-like modes that are typical of thin-walled structures that cannot be detected using classical beam models. The refined 1D models used in the present work provide results that are as accurate as those from solid FE models, but the 1D models have a much lower computational cost.

• Imaging Science in Dentistry
• /
• v.49 no.4
• /
• pp.257-263
• /
• 2019

Purpose: This study was performed to evaluate the dimensional accuracy of digital dental models constructed from cone-beam computed tomographic (CBCT) scans of polyvinyl siloxane (PVS) impressions and cast scan models. Materials and Methods: A pair of PVS impressions was obtained from 20 subjects and scanned using CBCT (resolution, 0.1 mm). A cast scan model was constructed by scanning the gypsum model using a model scanner. After reconstruction of the digital models, the mesio-distal width of each tooth, inter-canine width, and inter-molar width were measured, and the Bolton ratios were calculated and compared. The 2 models were superimposed and the difference between the models was measured using 3-dimensional analysis. Results: The range of mean error between the cast scan model and the CBCT scan model was -0.15 mm to 0.13 mm in the mesio-distal width of the teeth and 0.03 mm to 0.42 mm in the width analysis. The differences in the Bolton ratios between the cast scan models and CBCT scan models were 0.87 (anterior ratio) and 0.72 (overall ratio), with no significant difference (P>0.05). The mean maxillary and mandibular difference when the cast scan model and the CBCT scan model were superimposed was 53 ㎛. Conclusion: There was no statistically significant difference in most of the measurements. The maximum tooth size difference was 0.15mm, and the average difference in model overlap was 53 ㎛. Digital models produced by scanning impressions at a high resolution using CBCT can be used in clinical practice.

• Journal of Korean Society for Atmospheric Environment
• /
• v.13 no.1
• /
• pp.79-89
• /
• 1997

In order to reduce the outbreaks of short-term high concentrations and its impacts, we developed the models which predicted tomorrow's maximum hourly concentrations of $O_3$, TSP, SO$_2$, NO$_2$ and CO. Statistical methods like multi regressions were used because it must be operated easily under the present conditions. 47 independent variables were used, which included observed concentrations of air pollutants, observed and forcasted meteorological data in 1994 at Seoul and its surrounding areas. We subdivided Seoul into 4 areas coinciding with the present ozone warning areas. 4 kinds of seasonal models were developed due to the seasonal variations of observed concentrations, and 2 kinds of data models for the unavailable case of forecasted meteorological data. By comparing the $R^2$and root mean square error(hearafter 'RMSE') of each model, we confirmed that the models including forecasted data showed higher accuracy than ones using observed only. It was also shown that the higher the seasonal mean concentrations, the larger the RMSE. There was no distinct difference between the results of 4 areal models. In case of test run using 1995's data, the models predicted well the trends of daily variation of concentrations and the days when the possibility of outbreak of high concentarion was high. This study showed that it was reasonable to use those models as operational ones, because the $R^2$ and RMSE of models were smaller than those of operational/research models such as in South Coast Air Basin, CA, USA.

• Journal of The Korean Association For Science Education
• /
• v.31 no.3
• /
• pp.475-489
• /
• 2011

The purposes of this study were to inform the exemplary models of integrated science and mathematics and to analyze and discuss their similarities and differences of the models. There were two steps to select the exemplary models of integrated science and mathematics. First, the second volume (Berlin & Lee, 2003) of the bibliography of integrated science and mathematics was analyzed to identify the models. As a second step, we selected the models that are dealt with in the School Science Mathematics journal and were cited more than three times. The findings showed that the following four exemplary theoretical models were identified and published in the SSM journal: the Berlin-White Integrated Science and Mathematics (BWISM) Model, the Mathematics/Science Continuum Model, the Continuum Model of Integration, and the Five Types of Science and Mathematics Integration. The Berlin-White Integrated Science and Mathematics (BWISM) Model focused an interpretive or framework theory for integrated science and mathematics teaching and learning. BWISM focused on a conceptual base and a common language for integrated science and mathematics teaching and learning. The Mathematics/Science Continuum Model provided five categories and ways to clarify the extent of overlap or coordination between science and mathematics during instructional practice. The Continuum Model of Integration included five categories and clarified the nature of the relationship between the mathematics and science being taught and the curricular goals for the disciplines. These five types of science and mathematics integrations described the method, type, and instructional implications of five different approaches to integration. The five categories focused on clarifying various forms of integrated science and mathematics education. Several differences and similarities among the models were identified on the basis of the analysis of the content and characteristics of the models. Theoretically, there is strong support for the integration of science and mathematics education as a way to enhance science and mathematics learning experiences. It is expected that these instructional models for integration of science and mathematics could be used to develop and evaluate integration programs and to disseminate integration approaches to curriculum and instruction.

• Journal of the Korean Institute of Illuminating and Electrical Installation Engineers
• /
• v.16 no.2
• /
• pp.98-103
• /
• 2002

In case of power demand forecasting, the most important problems are to deal with the load of special-days. Accordingly, this paper presents the method that forecasting long (the Lunar New Year, the Full Moon Festival) and short(the Planting Trees Day, the Memorial Day, etc) special-days peak load using neural networks and regression models. long and short special-days peak load forecast by neural networks models uses pattern conversion ratio and four-order orthogonal polynomials regression models. There are using that special-days peak load data during ten years(1985∼1994). In the result of special-days peak load forecasting, forecasting % error shows good results as about 1 ∼2[%] both neural networks models and four-order orthogonal polynomials regression models. Besides, from the result of analysis of adjusted coefficient of determination and F-test, the significance of the are convinced four-order orthogonal polynomials regression models. When the neural networks models are compared with the four-order orthogonal polynomials regression models at a view of the results of special-days peak load forecasting, the neural networks models which uses pattern conversion ratio are more effective on forecasting long special-days peak load. On the other hand, in case of forecasting short special-days peak load, both are valid.

• Korean journal of applied entomology
• /
• v.56 no.1
• /
• pp.1-18
• /
• 2017

Temperature-dependent development model is an essential component for forecasting models of insect pests as well as for insect population models. This study reviewed the nonlinear models which explain the relationship between temperature and development rate of insects. In the present study, the types of models were classified largely into empirical and biophysical model, and the groups were subdivided into subgroups according to the similarity of mathematical equations or the connection with original idea. Empirical models that apply analytical functions describing the suitable shape of development curve were subdivided into multiple subgroups as Stinner-based types, Logan-based types, performance models and Beta distribution types. Biophysical models based on enzyme kinetic reaction were grouped as monophyletic group leading to Eyring-model, SM-model, SS-mode, and SSI-model. Finally, we described the historical development and characteristics of non-linear development models and discussed the availability of models.

• Spatial Information Research
• /
• v.21 no.2
• /
• pp.45-54
• /
• 2013

Analytical and simulation error models have the ability to describe (or realize) error-corrupted versions of spatial data. But the different approaches for modeling positional errors require an internal validation that ascertains whether the analytical and simulation error models predict correct positional errors in a defined set of conditions. This paper presents stochastic simulation models of a point and a line segm ent to be validated w ith analytical error models, which are an error ellipse and an error band model, respectively. The simulation error models populate positional errors by the Monte Carlo simulation, according to an assumed error distribution prescribed by given parameters of a variance-covariance matrix. In the validation process, a set of positional errors by the simulation models is compared to a theoretical description by the analytical error models. Results show that the proposed simulation models realize positional uncertainties of the same spatial data according to a defined level of positional quality.