• Journal of Intelligence and Information Systems
• /
• v.17 no.4
• /
• pp.241-254
• /
• 2011

Financial time-series forecasting is one of the most important issues because it is essential for the risk management of financial institutions. Therefore, researchers have tried to forecast financial time-series using various data mining techniques such as regression, artificial neural networks, decision trees, k-nearest neighbor etc. Recently, support vector machines (SVMs) are popularly applied to this research area because they have advantages that they don't require huge training data and have low possibility of overfitting. However, a user must determine several design factors by heuristics in order to use SVM. For example, the selection of appropriate kernel function and its parameters and proper feature subset selection are major design factors of SVM. Other than these factors, the proper selection of instance subset may also improve the forecasting performance of SVM by eliminating irrelevant and distorting training instances. Nonetheless, there have been few studies that have applied instance selection to SVM, especially in the domain of stock market prediction. Instance selection tries to choose proper instance subsets from original training data. It may be considered as a method of knowledge refinement and it maintains the instance-base. This study proposes the novel instance selection algorithm for SVMs. The proposed technique in this study uses genetic algorithm (GA) to optimize instance selection process with parameter optimization simultaneously. We call the model as ISVM (SVM with Instance selection) in this study. Experiments on stock market data are implemented using ISVM. In this study, the GA searches for optimal or near-optimal values of kernel parameters and relevant instances for SVMs. This study needs two sets of parameters in chromosomes in GA setting : The codes for kernel parameters and for instance selection. For the controlling parameters of the GA search, the population size is set at 50 organisms and the value of the crossover rate is set at 0.7 while the mutation rate is 0.1. As the stopping condition, 50 generations are permitted. The application data used in this study consists of technical indicators and the direction of change in the daily Korea stock price index (KOSPI). The total number of samples is 2218 trading days. We separate the whole data into three subsets as training, test, hold-out data set. The number of data in each subset is 1056, 581, 581 respectively. This study compares ISVM to several comparative models including logistic regression (logit), backpropagation neural networks (ANN), nearest neighbor (1-NN), conventional SVM (SVM) and SVM with the optimized parameters (PSVM). In especial, PSVM uses optimized kernel parameters by the genetic algorithm. The experimental results show that ISVM outperforms 1-NN by 15.32%, ANN by 6.89%, Logit and SVM by 5.34%, and PSVM by 4.82% for the holdout data. For ISVM, only 556 data from 1056 original training data are used to produce the result. In addition, the two-sample test for proportions is used to examine whether ISVM significantly outperforms other comparative models. The results indicate that ISVM outperforms ANN and 1-NN at the 1% statistical significance level. In addition, ISVM performs better than Logit, SVM and PSVM at the 5% statistical significance level.

• Journal of The Korean Association For Science Education
• /
• v.37 no.6
• /
• pp.921-935
• /
• 2017

The purpose of this research is to investigate trends of science educational researches on underrepresented students by scrutinizing Korean science education research literatures. For this particular purpose, literatures on underrepresented students were extracted from both listed and candidate journals for KCI and theses from 1984 to February 2017, and analyzed criteria such as source, year of publication, design, method, and content of research. A total of 125 papers from journals and 147 theses were extracted. In these researches, 61%, 20%, 6% were about students with disability, underachievers, and North Korean defector students respectively. The ratio of the researches on other underrepresented students such as multicultural, low income families, students who are from rural areas, and other underrepresented students were less than 5%. According to the year of publication, it was found that the number of research papers on underrepresented students increased continuously by a single digit from 1984 by focusing on the students with disability and underachievers. After that, from around 2008, it showed a rapid increase and researches on underrepresented students carried out more than 20 times annually. With regards to research design, there were 58% quantitative, 28% qualitative and 14% hybrid research design. Through analysis of research methods, we found that 30% of experimental research, 22% of interpretive research, 20% of correlation analysis, and 14% of survey research. After going through the characteristics of the research contents by visualizing the relationship between the research groups and the keywords that were extracted, it was found that even though the science education researches on underrepresented students have various contents, there were no keywords that were researched continuously and intensively in this area. The structural relationship between the keywords and each research group on underrepresented students showed that 'academic achievement' is the keyword with the highest degree of mediateness and connectedness.

• Restorative Dentistry and Endodontics
• /
• v.27 no.2
• /
• pp.142-149
• /
• 2002

Low-viscosity composite resins may produce better sealed margins than stiffer compositions (KempScholte and Davidson, 1988: Crim, 1989). Plowable composites have been recommended for use in Class V cavities but it is also controversial because of its high rates of shrinkage. On the other hand, in the study comparing elastic moduli and leakage, the microfill had the least leakage (Rundle et at. 1997) Furthermore, in the 1996 survey of the Reality Editorial Team, microfills were the clear choice for abfraction lesions. The purpose of this study was to evaluate the microleakage of 6 compostite resins (2 hybrids, 2 microfills, and 2 flowable composites) with and without load cycling. Notch-shaped Class V cavities were prepared on buccal surface of 180 extracted human upper premolars on cementum margin. The teeth were randomly divided into non-load cycling group (group 1) and load cycling group (group 2) of 90 teeth each. The experimental teeth of each group were randomly divided into 6 subgroups of 15 samples. All preparations were etched, and Single bond was applied. Preparations were restored with the following materials (n=15) : hybrid composite resin [Z250(3M Dental Products Inc. St. Paul, USA), Denfil(Vericom, Ahnyang, Korea)], microfill [Heliomolar RO(Vivadent, Schaan, Liechtenstein), Micronew(Bisco Inc. Schaumburg, IL, USA)], and flowable composite［AeliteFlo(Bisco Inc. Schaumburg, IL, USA), Revolution(Kerr Corp. Orange, CA, USA)]. Teeth of group 2 were subjected to occlusal load (100N for 50,000 cycles) using chewing simulator(MTS 858 Mini Bionix II system, MTS Systems Corp. Minn. USA). All samples were coated with nail polish 1mm short of the restoration, placed in 2％ methylene blue for 24 hours, and sectioned with a diamond wheel. Enamel and dentin/cementum margins were analyzed for microleakage on a sclale of 0 (no leakage) to 3 (3/3 of wall). Results were statistically analyzed by Kruscal-Wallis One way analysis, Mann-Whitney U-test, and Student-Newmann-Keuls method. (p = 0.05) Results : 1. There was significantly less microleage in enamel margins than dentinal margins of all groups (p＜0.05) 2. There was no significant between six composite resin in enamel margin of group 1. 3. In dentin margin of group 1, flowable composite had more microleakage than others but not of significant differences. 4. there was no significant difference between six composite resin in enamel margin of group 2. 5. In dentin margin of group 2, the microleakage were R＞A =H=M＞D＞Z. But there was no significant differences. 6. In enamel margins, load cycling did not affect the marginal microleakage in significant degree. 7. In enamel margins, load cycling did affect the marginal microleakage only in Revolution. (p＜0.05).