• Journal of Intelligence and Information Systems
• /
• v.25 no.3
• /
• pp.239-251
• /
• 2019

Statistical methods such as moving averages, Kalman filtering, exponential smoothing, regression analysis, and ARIMA (autoregressive integrated moving average) have been used for stock market predictions. However, these statistical methods have not produced superior performances. In recent years, machine learning techniques have been widely used in stock market predictions, including artificial neural network, SVM, and genetic algorithm. In particular, a case-based reasoning method, known as k-nearest neighbor is also widely used for stock price prediction. Case based reasoning retrieves several similar cases from previous cases when a new problem occurs, and combines the class labels of similar cases to create a classification for the new problem. However, case based reasoning has some problems. First, case based reasoning has a tendency to search for a fixed number of neighbors in the observation space and always selects the same number of neighbors rather than the best similar neighbors for the target case. So, case based reasoning may have to take into account more cases even when there are fewer cases applicable depending on the subject. Second, case based reasoning may select neighbors that are far away from the target case. Thus, case based reasoning does not guarantee an optimal pseudo-neighborhood for various target cases, and the predictability can be degraded due to a deviation from the desired similar neighbor. This paper examines how the size of learning data affects stock price predictability through k-nearest neighbor and compares the predictability of k-nearest neighbor with the random walk model according to the size of the learning data and the number of neighbors. In this study, Samsung electronics stock prices were predicted by dividing the learning dataset into two types. For the prediction of next day's closing price, we used four variables: opening value, daily high, daily low, and daily close. In the first experiment, data from January 1, 2000 to December 31, 2017 were used for the learning process. In the second experiment, data from January 1, 2015 to December 31, 2017 were used for the learning process. The test data is from January 1, 2018 to August 31, 2018 for both experiments. We compared the performance of k-NN with the random walk model using the two learning dataset. The mean absolute percentage error (MAPE) was 1.3497 for the random walk model and 1.3570 for the k-NN for the first experiment when the learning data was small. However, the mean absolute percentage error (MAPE) for the random walk model was 1.3497 and the k-NN was 1.2928 for the second experiment when the learning data was large. These results show that the prediction power when more learning data are used is higher than when less learning data are used. Also, this paper shows that k-NN generally produces a better predictive power than random walk model for larger learning datasets and does not when the learning dataset is relatively small. Future studies need to consider macroeconomic variables related to stock price forecasting including opening price, low price, high price, and closing price. Also, to produce better results, it is recommended that the k-nearest neighbor needs to find nearest neighbors using the second step filtering method considering fundamental economic variables as well as a sufficient amount of learning data.

• Journal of the Korean Society of Urban Environment
• /
• v.18 no.4
• /
• pp.473-483
• /
• 2018

In this study, we have developed a lab scale bioreactor to identify the characteristics of nitrification reaction according to operation condition (temperature, inhibitor (as Cl), activated nitrifying bacteria (ANB). etc) to improve nitrification efficiency at low temperature. Recovery rate of nitrification took about 4 days to reach the normal level by injected ANB after inhibition shock of CI injection at $20^{\circ}C$, when measured the concentration of $NO_2{^-}-N+NO_3{^-}-N$ in the effluent. In the case of $10^{\circ}C$, recovery of nitrification rate took about 4 days to reach the level of half to the normal level and 7 days for complete recovery which took 3 days more than those at $20^{\circ}C$. At $10^{\circ}C$ considering the winter season, the specific nitrification rate(SNR) of the from 1 day to 6 days after injected ANB according to its operation condition increased from 0.029 to 0.767 mgN/gSS/hr. The simulated SNR for the 8th day after the injected ANB at $10^{\circ}C$ was 0.840, 3.625 mgN/gSS/hr, respectively as linear function and exponential function, expecting to exceed level of 2.592 mgN/gSS/hr at normal condition. It was confirmed that injection of ANB during low temperature operation has many effects for improving nitrification efficiency through this study. In future studies, if further studies are carried out the determination of ANB injection and the design of efficient ANB reactor considering the changes of operating characteristics by site, it will contribute to the improvement of nitrification efficiency in winter season.

• Journal of the Korea Organic Resources Recycling Association
• /
• v.30 no.4
• /
• pp.67-83
• /
• 2022

Biochar is a carbon material produced through the pyrolysis of agricultural biomass with limited oxygen condition. It has been suggested to enhance the carbon sequestration and mineralization of soil carbon. Objective of this study was to investigate soil potential carbon mineralization and carbon dioxide(CO₂) emissions in different soils cooperated with barely straw and livestock manure biochars in the closed chamber. The incubation was conducted during 49 days using a closed chamber. The treatments consisted of 2 different biochars that were originated from barley straw and livestock manure, and application amounts were 0, 5, 10 and 20 ton ha^-1 with different soils as upland, protected cultivation, converted and reclaimed. The results indicated that the TC increased significantly in all soils after biochar application. Mineralization of soil carbon was well fitted for Kinetic first-order exponential rate model equation (P<0.001). Potential mineralization rate ranged from 8.7 to 15.5% and 8.2 to 16.5% in the barely straw biochar and livestock manure biochar treatments, respectively. The highest CO₂ emission was 81.94 mg kg^-1 in the upland soil, and it was more emitted CO₂ for barely straw biochar application than its livestock biochar regardless of their application rates. Soil amendment of biochar is suitable for barely straw biochar regardless of application rates for mitigation of CO₂ emission in the cropland.

• Journal of Korean Society of Coastal and Ocean Engineers
• /
• v.34 no.2
• /
• pp.37-45
• /
• 2022

A simplified model using the lifetime distribution has been presented to estimate the Mean Residual Life (MRL) of rubble-mound breakwaters, which is not like a stochastic process model based on time-dependent history data to the cumulative damage progress of rubble-mound breakwaters. The parameters involved in the lifetime distribution can be easily estimated by using the upper and lower limits of lifetime and their likelihood that made a judgement by several experts taking account of the initial design lifetime, the past sequences of loads, and others. The simplified model presented in this paper has been applied to the rubble-mound breakwater with TTP armor layer. Wiener Process (WP)-based stochastic model also has been applied together with Monte-Carlo Simulation (MCS) technique to the breakwater of the same condition having time-dependent cumulative damage to TTP armor layer. From the comparison of lifetime distribution obtained from each models including Mean Time To Failure (MTTF), it has found that the lifetime distributions of rubble-mound breakwater can be very satisfactorily fitted by log-normal distribution for all types of cumulative damage progresses, such as exponential, linear, and logarithmic deterioration which are feasible in the real situations. Finally, the MRL of rubble-mound breakwaters estimated by the simplified model presented in this paper have been compared with those by WP stochastic process. It can be shown that results of the presented simplified model have been identical with those of WP stochastic process until any ages in the range of MTT F regardless of the deterioration types. However, a little of differences have been seen at the ages in the neighborhood of MTTF, specially, for the linear and logarithmic deterioration of cumulative damages. For the accurate estimation of MRL of harbor structures, it may be desirable that the stochastic processes should be used to consider properly time-dependent uncertainties of damage deterioration. Nevertheless, the simplified model presented in this paper can be useful in the building of the MRL-based preventive maintenance planning for several kinds of harbor structures, because of which is not needed time-dependent history data about the damage deterioration of structures as mentioned above.

• KOREAN JOURNAL OF CROP SCIENCE
• /
• v.30 no.3
• /
• pp.252-258
• /
• 1985

An experiment was conducted at the Yeongnam Crop Experiment Station to obtain basic informations about cultural techniques for high yielding by manipulating plant spacing using two rice cultivars, Samgangbyeo (Indica/Japonica type) and Nakdongbyeo (Japonica type), and four plant spacings, 10${\times}$10cm, 20${\times}$20cm 30${\times}$30cm and 40${\times}$40cm, with 4 kinds of seedling number per hill, 1,3,5 and 7, respectively. High photosynthetic efficiency (Eu) exhibited at the Samgangbyeo compared to Nakdongbyeo regardless of plant spacings and seedling numbers. For Samgangbyeo, Eu value was the highest at the 20${\times}$20cm plant spacing and five seedlings and seven seedlings per hill showed high Eu values at 10${\times}$10cm plant spacing and 20${\times}$20cm plant spacing, respectively, while other plant spacings were not significantly differed among seedling numbers. For Nakdongbyeo, however, one seedling plot obtained high Eu value at the 10${\times}$10cm plant spacing while this Eu value increased as the seedling number per hill increased in other plant spacings. There was a high positive correlation between rice grain yield and total competition index for both cultivars while kind of relationships differed in these two cultivars; linear relationship for Samgangbyeo and exponential relationship for Nakdongbyeo, respectively. Competition index between rice hill was more significant than within rice hill for Samgangbyeo while both competition indexs were important for Nakdongbyeo to increase rice yield.

• Journal of Intelligence and Information Systems
• /
• v.28 no.1
• /
• pp.69-88
• /
• 2022

One of the characteristics of South Korea's economic structure is that it is highly dependent on exports. Thus, many businesses are closely related to the global economy and diplomatic situation. In addition, small and medium-sized enterprises(SMEs) specialized in exporting are struggling due to the spread of COVID-19. Therefore, this study aimed to develop a model to forecast exports for next year to support SMEs' export strategy and decision making. Also, this study proposed a strategy to recommend promising export countries of each item based on the forecasting model. We analyzed important variables used in previous studies such as country-specific, item-specific, and macro-economic variables and collected those variables to train our prediction model. Next, through the exploratory data analysis(EDA) it was found that exports, which is a target variable, have a highly skewed distribution. To deal with this issue and improve predictive performance, we suggest a separated learning method. In a separated learning method, the whole dataset is divided into homogeneous subgroups and a prediction algorithm is applied to each group. Thus, characteristics of each group can be more precisely trained using different input variables and algorithms. In this study, we divided the dataset into five subgroups based on the exports to decrease skewness of the target variable. After the separation, we found that each group has different characteristics in countries and goods. For example, In Group 1, most of the exporting countries are developing countries and the majority of exporting goods are low value products such as glass and prints. On the other hand, major exporting countries of South Korea such as China, USA, and Vietnam are included in Group 4 and Group 5 and most exporting goods in these groups are high value products. Then we used LightGBM(LGBM) and Exponential Moving Average(EMA) for prediction. Considering the characteristics of each group, models were built using LGBM for Group 1 to 4 and EMA for Group 5. To evaluate the performance of the model, we compare different model structures and algorithms. As a result, it was found that the separated learning model had best performance compared to other models. After the model was built, we also provided variable importance of each group using SHAP-value to add explainability of our model. Based on the prediction model, we proposed a second-stage recommendation strategy for potential export countries. In the first phase, BCG matrix was used to find Star and Question Mark markets that are expected to grow rapidly. In the second phase, we calculated scores for each country and recommendations were made according to ranking. Using this recommendation framework, potential export countries were selected and information about those countries for each item was presented. There are several implications of this study. First of all, most of the preceding studies have conducted research on the specific situation or country. However, this study use various variables and develops a machine learning model for a wide range of countries and items. Second, as to our knowledge, it is the first attempt to adopt a separated learning method for exports prediction. By separating the dataset into 5 homogeneous subgroups, we could enhance the predictive performance of the model. Also, more detailed explanation of models by group is provided using SHAP values. Lastly, this study has several practical implications. There are some platforms which serve trade information including KOTRA, but most of them are based on past data. Therefore, it is not easy for companies to predict future trends. By utilizing the model and recommendation strategy in this research, trade related services in each platform can be improved so that companies including SMEs can fully utilize the service when making strategies and decisions for exports.

• Economic and Environmental Geology
• /
• v.56 no.6
• /
• pp.781-797
• /
• 2023

Recently, there has been a rapid response from mineral-demanding countries for securing critical minerals in a high tech industries. Graphite, while overwhelmingly dominated by China in production, is changing in global supply due to the exponential growth in EV battery sector, with active exploration in East Africa. Rare earth elements are essential raw materials widely used in advanced industries. Globally, there are ongoing developments in the production of REEs from three main deposit types: carbonatite, laterite, and ion-adsorption clay types. While China's production has decreased somewhat, it still maintains overwhelming dominance in this sector. Recent changes over the past few years include the rapid emergence of Myanmar and increased production in Vietnam. Nickel has been used in various chemical and metal industries for a long time, but recently, its significance in the market has been increasing, particularly in the battery sector. Worldwide, nickel deposits can be broadly classified into two types: laterite-type, which are derived from ultramafic rocks, and ultramafic hosted sulfide-type. It is predicted that the development of sulfide-type, primarily in Australia, will continue to grow, while the development of laterite-type is expected to be promoted in Indonesia. This is largely driven by the growing demand for nickel in response to the demand for lithium-ion batteries. The global lithium ores are produced in three main types: brine lake (78%), rock/mineral (19%), and clay types (3%). Rock/mineral type has a slightly higher grade compared to brine lake type, but they are less abundant. Chile, Argentina, and the United States primarily produce lithium from brine lake deposits, while Australia and China extract lithium from both brine lake and rock/mineral sources. Canada, on the other hand, exclusively produces lithium from rock/mineral type. Vanadium has traditionally been used in steel alloys, accounting for approximately 90% of its usage. However, there is a growing trend in the use for vanadium redox flow batteries, particularly for large-scale energy storage applications. The global sources of vanadium can be broadly categorized into two main types: vanadium contained in iron ore (81%) produced from mines and vanadium recovered from by-products (secondary sources, 18%). The primary source, accounting for 81%, is vanadium-iron ores, with 70% derived from vanadium slag in the steel making process and 30% from ore mined in primary sources. Intermediate vanadium oxides are manufactured from these sources. Vanadium deposits are classified into four types: vanadiferous titanomagnetite (VTM), sandstone-hosted, shale-hosted, and vanadate types. Currently, only the VTM-type ore is being produced.

• KOREAN JOURNAL OF CROP SCIENCE
• /
• v.68 no.4
• /
• pp.236-245
• /
• 2023

We investigated the basic vegetative phase (BVP), photoperiod-sensitivity (PS), and thermo-sensitivity (TS) of 40 rice varieties to characterize their heading ecology that is adaptable to North Korean rice production areas. The ranges were 12 - 43 days for BVP, 0 - 74 days for PS, and 15 - 33 days for TS, depending on the variety. The number of days from transplanting to the heading stage (DTH) was significantly correlated with PS in the 13 major rice production regions where all 40 varieties (including early-, middle-, and mid-late-maturing varieties) were tested. DTH was significantly correlated with BVP and TS but not with PS in the 10 low-temperature regions where only 28 early-maturing varieties were tested. The heading ecology of the adaptable varieties for each region could be characterized by an exponential equation between the BVP and PS of varieties that headed at the border of the safe marginal heading date (SMHD) for each of the seven regional environments (Kaesong, Haeju, Yongyon, Singye, Sariwon, Nampo, and Pyongyang). A PS of 25 - 30 days or less was an additional adaptive trait in the Sariwon and Pyongyang environments. The varieties that reached the heading stage not later than the SMHD in six regional environments (Anju, Kusong, Sinuiju, Changjon, Wonsan, and Supung) and those that reached the heading stage not later than the late marginal heading date (LMHD) in four regional environments (Hamhung, Pyonggang, Huichon, and Kanggye) had both a PS of 26 days or less and a BVP of 25 - 28 days or less. In the Yangdok, Sinpo, and Chunggang environments, varieties that reached the heading stage not later than the LMHD for each region had a BVP of 15 - 20 days or less. The results suggested that a shortened BVP trait should be introduced to existing early-maturing rice varieties to reduce the duration of growth period to reach the heading stage.

• Journal of Aquaculture
• /
• v.21 no.4
• /
• pp.203-212
• /
• 2008

A nutritional demand of oyster, Crassostrea gigas larva as part of research for improving of utilization of microalgae being used for the artificial oyster seed production. The change of body growth and biochemical compositions of larvae were investigated during larvae rearing in hatchery. The larvae were cultured in 60 M/T tank and fed mixture 6 different phytoplankton species, Isochrysis galbana (30%), Cheatoceros gracilis (20%), Pavlova lutheri (20%), Phaeodactylum triconutum (10%), Nannochryis oculata (10%) and Tetraselmis tetrathele (10%). The initial feeding amount was $0.3{\times}10^4cells/mL$ at three times a day to D-shaped larva and the feeding amount had been increased 30% gradually every two day since the larvae were raising. The larvae were developed from D shape to pediveliger stage for 12 days. The daily growth of shell length and hight were $5.8{\sim}30.8\;{\mu}m$ and $8.7{\sim}31.4\;{\mu}m$, respectively and weight gains were changed from D shape to pediveliger as follow: wet weight was $0.52{\sim}15.0\;{\mu}g/larva$, dry weight was $0.2{\sim}6.5\;{\mu}g/larva$, and ash free dry weight was $0.1{\sim}8.5\;{\mu}g/larva$. The larvae growth pattern shown a logarithmic phase from D shape to umbone stage but after that stage shown a exponential growth aspect. The crude protein, crude lipid and nitrogen free extract (NFE) of larvae during rearing periods were analyzed as $6.1{\sim}10.6%$, $0.6{\sim}1.1%$ and 1.0-2.7%, respectively. And the total amino acid contents of the larvae during rearing periods were in order as glutamic acid $1.26{\sim}2.24%$, aspartic acid $0.97{\sim}1.70%$, and methionine $0.12{\sim}0.33%$. Of the total fatty acid in the analyzed larvae, the saturated fatty acid (SSAFA) was decreased from 54.3% (D shaped larvae) to 17.1 % (pediveliger) as larvae development but the total mono-unsaturated fatty acid (${\Sigma}MOFA$) and Poly-unsaturated fatty acid (${\Sigma}PUFA$) were increased from 29.9% and 7.8% to 40.6% and 45.6%, respectively. By the way the each fatty acid of the larvae were composed of palmitic acid $9.89{\sim}36.95%$, oleic acid $12.17{\sim}32.29%$, linoleic acid $1.96{\sim}33.55%$, EPA $2.17{\sim}11.58%$ and DHA $1.95{\sim}4.51%$. As a result of this study, the larvae of oyster were demanded a various nutrients for healthy growth and the feeding control, expecially after umbone stage larvae are a rapidly growing time, is very important for success of artificial seed production.

• Korean Journal of Mineralogy and Petrology
• /
• v.33 no.3
• /
• pp.273-291
• /
• 2020

The strength characteristics of the three orthogonal splitting planes, known as rift, grain and hardway planes in granite quarries, were examined. R, G and H specimens were obtained from the block samples of Jurassic granites in Geochang and Hapcheon areas. The directions of the long axes of these three specimens are perpendicular to each of the three planes. First, The chart, showing the scaling characteristics of three graphs related to the uniaxial compressive strengths of R, G and H specimens, were made. The graphs for the three specimens, along with the increase of strength, are arranged in the order of H < G < R. The angles of inclination of the graphs for the three specimens, suggesting the degree of uniformity of the texture within the specimen, were compared. The above angles for H specimens(θ^H, 24.0°~37.3°) are the lowest among the three specimens. Second, the scaling characteristics related to the three graphs of RG, GH and RH specimens, representing a combination of the mean compressive strengths of the two specimens, were derived. These three graphs, taking the various N-shaped forms, are arranged in the order of GH < RH < RG. Third, the correlation chart between the strength difference(Δσ_t) and the angle of inclination(θ) was made. The above two parameters show the correlation of the exponential function with an exponent(λ) of -0.003. In both granites, the angle of inclination(θ^RH) of the RH-graph is the lowest. Fourth, the six types of charts, showing the correlations among the three kinds of compressive strengths for the three specimens and the five parameters for the two sets of microcracks aligned parallel to the compressive load applied to each specimen, were made. From these charts for Geochang and Hapcheon granites, the mean value(0.877) of the correlation coefficients(R²) for total density(Lt), along with the frequency(N, 0.872) and density(ρ, 0.874), is the highest. In addition, the mean values(0.829) of correlation coefficients associated with the mean compressive strengths are more higher than the minimum(0.768) and maximum(0.804) compression strengths of three specimens. Fifth, the distributional characteristics of the Brazilian tensile strengths measured in directions parallel to the above two sets of microcracks in the three specimens from Geochang granite were derived. From the related chart, the three graphs for these tensile strengths corresponding to the R, G and H specimens show an order of H(R1+G1) < G(R2+H1) < R(R1+G1). The order of arrangement of the three graphs for the tensile strengths and that for the compressive strengths are mutually consistent. Therefore, the compressive strengths of the three specimens are proportional to the three types of tensile strengths. Sixth, the values of correlation coefficients, among the three tensile strengths corresponding to each cumulative number(N=1~10) from the above three graphs and the five parameters corresponding to each graph, were derived. The mean values of correlation coefficients for each parameter from the 10 correlation charts increase in the order of density(0.763) < total length(0.817) < frequency(0.839) < mean length(Lm, 0.901) ≤ median length(Lmed, 0.903). Seventh, the correlation charts among the compressive strengths and tensile strengths for the three specimens were made. The above correlation charts were divided into nine types based on the three kinds of compressive strengths and the five groups(A~E) of tensile strengths. From the related charts, as the tensile strength increases with the mean and maximum compressive strengths excluding the minimum compressive strength, the value of correlation coefficient increases rapidly.