• Journal of Intelligence and Information Systems
• /
• v.26 no.1
• /
• pp.23-45
• /
• 2020

Big data is creating in a wide variety of fields such as medical care, manufacturing, logistics, sales site, SNS, and the dataset characteristics are also diverse. In order to secure the competitiveness of companies, it is necessary to improve decision-making capacity using a classification algorithm. However, most of them do not have sufficient knowledge on what kind of classification algorithm is appropriate for a specific problem area. In other words, determining which classification algorithm is appropriate depending on the characteristics of the dataset was has been a task that required expertise and effort. This is because the relationship between the characteristics of datasets (called meta-features) and the performance of classification algorithms has not been fully understood. Moreover, there has been little research on meta-features reflecting the characteristics of multi-class. Therefore, the purpose of this study is to empirically analyze whether meta-features of multi-class datasets have a significant effect on the performance of classification algorithms. In this study, meta-features of multi-class datasets were identified into two factors, (the data structure and the data complexity,) and seven representative meta-features were selected. Among those, we included the Herfindahl-Hirschman Index (HHI), originally a market concentration measurement index, in the meta-features to replace IR(Imbalanced Ratio). Also, we developed a new index called Reverse ReLU Silhouette Score into the meta-feature set. Among the UCI Machine Learning Repository data, six representative datasets (Balance Scale, PageBlocks, Car Evaluation, User Knowledge-Modeling, Wine Quality(red), Contraceptive Method Choice) were selected. The class of each dataset was classified by using the classification algorithms (KNN, Logistic Regression, Nave Bayes, Random Forest, and SVM) selected in the study. For each dataset, we applied 10-fold cross validation method. 10% to 100% oversampling method is applied for each fold and meta-features of the dataset is measured. The meta-features selected are HHI, Number of Classes, Number of Features, Entropy, Reverse ReLU Silhouette Score, Nonlinearity of Linear Classifier, Hub Score. F1-score was selected as the dependent variable. As a result, the results of this study showed that the six meta-features including Reverse ReLU Silhouette Score and HHI proposed in this study have a significant effect on the classification performance. (1) The meta-features HHI proposed in this study was significant in the classification performance. (2) The number of variables has a significant effect on the classification performance, unlike the number of classes, but it has a positive effect. (3) The number of classes has a negative effect on the performance of classification. (4) Entropy has a significant effect on the performance of classification. (5) The Reverse ReLU Silhouette Score also significantly affects the classification performance at a significant level of 0.01. (6) The nonlinearity of linear classifiers has a significant negative effect on classification performance. In addition, the results of the analysis by the classification algorithms were also consistent. In the regression analysis by classification algorithm, Naïve Bayes algorithm does not have a significant effect on the number of variables unlike other classification algorithms. This study has two theoretical contributions: (1) two new meta-features (HHI, Reverse ReLU Silhouette score) was proved to be significant. (2) The effects of data characteristics on the performance of classification were investigated using meta-features. The practical contribution points (1) can be utilized in the development of classification algorithm recommendation system according to the characteristics of datasets. (2) Many data scientists are often testing by adjusting the parameters of the algorithm to find the optimal algorithm for the situation because the characteristics of the data are different. In this process, excessive waste of resources occurs due to hardware, cost, time, and manpower. This study is expected to be useful for machine learning, data mining researchers, practitioners, and machine learning-based system developers. The composition of this study consists of introduction, related research, research model, experiment, conclusion and discussion.

• Journal of the Korea Concrete Institute
• /
• v.20 no.1
• /
• pp.89-97
• /
• 2008

By replacing the meta-kaolin with cement and the waste tire chip with fine aggregate separately, the high strength concrete is protected from the spalling in fire and the method to constrain the temperature increase of steel bar within the concrete and the basic properties of the high strength concrete mixed with the material are reviewed. As the result, meta-kaolin increases the self fire proof characteristics of the concrete, the waste tire chip can share the internal expanding pressure so it can be deleted. In detail, using the meta-kaolin about the cement in 4$\sim$8% of weight ratio about the cement and the waste tire chip under the grade scope of 0.6$\sim$3 mm in 5$\sim$10% of weight ratio about the sand is very effective to prevent the spalling.

• Asian Pacific Journal of Cancer Prevention
• /
• v.14 no.4
• /
• pp.2337-2342
• /
• 2013

Meta-analyses have shown that microRNA polymorphisms have variable effects in different population. Yet, no meta-analysis investigated the association of two common polymorphisms of miRNA, mir-499 rs3746444 polymorphism and mir-149 rs2292832 polymorphism, with cancer risk in the Chinese population. We searched the PubMed, Web of Knowledge, MEDLINE, CNKI databases, as well as Cochrane library, updated on December 31, 2012 for assays regarding cancer risk association with these two common polymorphisms in the present meta-analysis. Odds ratios (ORs) and 95% confidence intervals (95% CIs) were used to explore the strength of associations. The results showed that rs3746444 polymorphism was associated with increased cancer risk (dominant model: GG/AG vs. AA: OR = 1.43, 95% CI: 1.14-1.80; recessive model: GG vs. AG/AA: OR = 1.54, 95% CI: 1.04-2.30; homozygote model: GG vs. AA: OR = 1.69, 95% CI: 1.10-2.60; heterozygote model: AG vs. AA: OR = 1. 35, 95% CI: 1.09-1.67), and rs3746444 was associated with liver cancer in the subgroup of cancer types. For the rs2292832 polymorphism, the results showed no significant risk association in both overall pooled analysis and subgroup of cancer types, smoking status, gender and tea drinking status in the Chinese population. This meta-analysis suggested that the rs3746444 GG genotype is associated with increased cancer risk, especially liver cancer, while the rs2292832 polymorphism showed no association with cancer risk in Chinese.

• Asian Pacific Journal of Cancer Prevention
• /
• v.14 no.4
• /
• pp.2407-2412
• /
• 2013

Background and Objectives: Data on associations between soy food intake after cancer diagnosis with breast cancer survival are conflicting, so we conducted this meta-analysis for more accurate evaluation. Methods: Comprehensive searches were conducted to find cohort studies of the relationship between soy food intake after cancer diagnosis and breast cancer survival. Data were analyzed with comprehensive meta-analysis software. Results: Five cohort studies (11,206 patients) were included. Pooling all comparisons, soy food intake after diagnosis was associated with reduced mortality (HR 0.85, 95%CI 0.77 0.93) and recurrence (HR 0.79, 95%CI 0.72 0.87). Pooling the comparisons of highest vs. lowest dose, soy food intake after diagnosis was again associated with reduced mortality (HR 0.84, 95%CI 0.71 0.99) and recurrence (HR 0.74, 95%CI 0.64 0.85). Subgroup analysis of ER status showed that soy food intake was associated with reduced mortality in both ER negative (highest vs. lowest: HR 0.75, 95%CI 0.64 0.88) and ER positive patients (highest vs. lowest: HR 0.72, 95%CI 0.61 0.84), and both premenopausal (highest vs. lowest: HR 0.78, 95%CI 0.69 0.88) and postmenopausal patients (highest vs. lowest: HR 0.81, 95%CI 0.73 0.91). In additioin, soy food intake was associated with reduced recurrence in ER negative (highest vs. lowest: HR 0.64, 95%CI 0.44 0.94) and ER+/PR+ (highest vs. lowest: HR 0.65, 95%CI 0.49 0.86), and postmenopausal patients (highest vs. lowest: HR 0.67, 95%CI 0.56 0.80). Conclusion: Our meta-analysis showed that soy food intake might be associated with better survival, especially for ER negative, ER+/PR+, and postmenopausal patients.

• Asian Pacific Journal of Cancer Prevention
• /
• v.14 no.4
• /
• pp.2515-2519
• /
• 2013

The standard surgery for early-stage endometrial cancer is total abdominal hysterectomy (TAH), while total laparoscopic hysterectomy (TLH) is less invasive and assumed to be associated with lower morbidity. This meta-analysis was performed to investigate the effects of TLH versus TAH in women with early-stage endometrial cancer. We searched the PubMed, EMBASE, CBM and Cochrane Review databases for randomized trials assessing the effects of TLH versus TAH in women with early-stage endometrial cancer. The relative risks (RR) with 95% confidence intervals (CIs) from each study were pooled using meta-analysis. In our study, 9 randomized trials with a total of 1,263 patients were included. Meta-analyses showed that TLH was associated with lower risks of major complications (RR = 0.53, 95%CI 0.29-0.98, P = 0.042), total complications (RR = 0.59, 95%CI 0.42-0.82, P = 0.002) and postoperative complications (RR = 0.57, 95%CI 0.40-0.83, P = 0.003). However, there were no obvious differences in risks of intra-operative complications (RR = 0.98, 95%CI 0.62-1.55, P = 0.919) and mortality (RR = 0.96, 95%CI 0.66-1.40, P = 0.835). In conclusion, our results provide new evidence of a benefit for TLH over TAH in terms of major complications, total complications and postoperative complications in endometrial cancer patients.

• Asian Pacific Journal of Cancer Prevention
• /
• v.14 no.4
• /
• pp.2413-2419
• /
• 2013

Background: Use of epidermal growth factor receptor inhibitors (EGFR-TKIs ) is now standard for non-small-cell lung cancer (NSCLC). However, the effects of EGFR-TKIs in maintenance therapy for advanced NSCLC patients are still unclear. The preent meta-analysis was performed to examine pooled data of randomized control trials (RCT) where EGFR-TKIs were compared against placebo in maintenance regimens for patients with advanced NCSLC to quantify potential benefits and determine safety. Methods: Several data bases were searched, including PubMed, EMBASE and CENTRAL, and we performed an internet search of conference literature. The endpoints were objective response rates (ORR), progression-free survival (PFS) and overall survival (OS). We performed a meta-analysis of the published data, using Comprehensive Meta Analysis software (Version 2.0). with a fixed effects model and an additional random effects model, when applicable. The results of the meta-analysis are expressed as hazard ratios (HRs) or risk ratios (RRs), with their corresponding 95% confidence intervals (95%CIs). Results: The final analysis included six trials, covering 3,758 patients. Compared with placebo, EGFR-TKIs maintenance therapy improved ORR and PFS for patients with advanced NSCLC, the difference being statistically significant (P<0.05), but proved unable to prolong patients' OS. The main adverse reactions were diarrhea and rashes. Conclusion: EGFR-TKIs demonstrated encouraging efficacy, safety and survival when delivered as maintenance therapy for patients with advanced NSCLC after first-line chemotherapy, especially for the patients who had adenocarcinomas, were female, non-smokers and patients with EGFR gene mutations.

• Journal of Korean Academy of Nursing
• /
• v.43 no.6
• /
• pp.746-759
• /
• 2013

Purpose: Infrared thermometers are increasingly used as a convenient, non-invasive assessment method for febrile children. However, the diagnostic accuracy of the infrared thermometer for children has been questioned, particularly in relation to sensitivity and specificity. The aim of this study was to evaluate diagnostic accuracy of infrared thermometers in febrile children. Methods: Articles published between 1966 and 2012 from periodicals indexed in the Ovid Medline, Embase, CINAHL, Cochrane, KoreaMed, NDSL, KERIS and other databases were selected, using the following keywords: 'infrared thermometer'. The QUADAS-II was applied to assess the internal validity of the diagnostic studies. Selected studies were analyzed using meta-analysis with MetaDisc 1.4. Results: Nineteen diagnostic studies with high methodological quality, involving 4,304 children, were included. The results of meta-analysis showed that the pooled sensitivity, specificity and AUC (Area Under the Curve) of infrared tympanic thermometers in children over 1 year were 0.80 (95% CI 0.78, 0.81), 0.94 (95% CI 0.93, 0.95) and 0.95 respectively. However the diagnostic accuracy of infrared tympanic thermometers in children with hyperthermia was low. Conclusion: The diagnostic accuracy of infrared tympanic thermometer was similar to axillary and rectal thermometers indicating a need for further research to substantiate these findings in children with hyperthermia.

• Korean Journal of Human Ecology
• /
• v.17 no.4
• /
• pp.785-796
• /
• 2008

The Open-Market internet shopping malls has rapidly been growing over the past few years. Thus, this study compares Meta-Mall and Open-Market, as well as evaluates image factors of the internet shopping mall in order to verify different levels of influence on customer satisfaction and revisit intentions. The subject of stud was made for students enrolled in colleges in the Busan area, using this Data collection method. Data were gathered using a questionnaire and through personal interviews from a sampling made from May 1 to May 8, 2008. The final 337 accomplished questionnaire forms were used for the subsequent analysis. The results of the substantial analysis are as follows. First, the following factors affect the image of internet shopping malls: assortment of goods and information service, site design, convenience, reliability of goods and information, and customer service. In Meta-Mall, these same factors, except customer service, were found to influence customer satisfaction, while in Open-Market, all the five factors were found to influence customer satisfaction. In Open-Market, convenience and reliability of goods and information had especially high levels of influence on customer satisfaction. Second, customer satisfaction with Internet shopping malls was found to influence the intentions of the customers to make a revisit. The result of the study shows that the factors of shopping mall image between Meta-Mall and Open-Market are perceived differently by their customers, and differences were found in the levels of their influence on customer satisfaction.

• Asian Pacific Journal of Cancer Prevention
• /
• v.15 no.6
• /
• pp.2571-2576
• /
• 2014

Background: Published studies on the association between the exonuclease 1 (EXO1) Glu589Lys polymorphism and cancer susceptibility have yielded conflicting results. Thus, a meta-analysis of published studies was performed to assess the possible association. Materials and Methods: All eligible case-control studies published up to January 2013 on the association between the EXO1 Glu589Lys polymorphism and cancer susceptibility were identified by searching PubMed, Web of Science, Science Direct and hand search. Either fixed-effect or random-effect models were used to calculate pooled odds ratios (ORs) with 95% confidence intervals (CIs) using the Comprehensive Meta-Analysis software version 2.2. Results: A total of 4,391 cancer cases and 4,339 controls from 10 studies were included. Overall, no significant association between the EXO1 Glu589Lys polymorphism and cancer susceptibility was observed in either genetic model. However; in subgroup analyses by cancer type, a significant association between EXO1 Glu589Lys and lung cancer risk was found (Lys vs Glu: OR=1.23, 95%CI=1.07-1.41, $p_{heterogeneity}$=0.05). Further, subgroup analysis by ethnicity indicated that there was a statistically increased cancer risk in Asians (Lys vs Glu: OR=1.42, 95%CI=1.30-1.55, $p_{heterogeneity}$=0.07; Lys/Lys vs Glu/Glu: OR=1.93, 95%CI=1.20-3.12, $p_{heterogeneity}$=0.01; Lys/Lys+Glu/Lys vs Glu/Glu: OR=1.52, 95%CI=1.37-1.68, $p_{heterogeneity}$=0.42; Lys/Lys vs Glu/Lys+Glu/Glu: OR=1.68, 95%CI=1.07-2.65, $p_{heterogeneity}$=0.02). However, significant association was absent in Caucasians. Conclusions: This meta-analysis suggests, for the first time, that the EXO1 Glu589Lys polymorphism is not associated with overall cancer susceptibility, although marginal associations were found for lung cancer and Asian subgroups. Additional well-designed studies with larger sample size focusing on different ethnicities and cancer types are needed to confirm these findings.

• Journal of Korean Critical Care Nursing
• /
• v.9 no.1
• /
• pp.51-65
• /
• 2016

Purpose: This study aimed to estimate the pooled prevalence of workplace bullying among nurses and examine the effect of moderate factors. Methods: Studies that assessed nursing workplace bullying using the negative act questionnaire developed by Einarsen et al. were searched for in 8 databases. The pooled prevalence was calculated with overall event rate and meta-ANOVA was performed for subgroup analysis. A funnel plot and Egger's linear regression asymmetry test were performed to confirm the publication bias of meta-analyzed studies. Results: A total of 23 studies (N=11,748) were included in the systematic review, and 14 studies (N=4,585) that reported prevalence using operational criteria of victim were selected for meta-analysis. The pooled prevalence of workplace bullying was 22.2% (95% CI: 19.0~25.7). Domestic studies reported lower prevalence than international studies (18.6% vs. 27.0%, Q=8.75, p=.003). The prevalences were different according to publication year (30.6% in 2008~2010, 19.5% in 2011~2014, 21.8% in 2015, Q=11.89, p=.003). Studies that recruited only new nurses reported higher prevalence than others (28.4% vs. 20.6%, Q=3.92, p=.048). Publication bias was not found. Conclusions: Workplace bullying is prevalent in the nursing profession, and is more serious for new graduate nurses. Individual and organizational measures for preventing workplace bullying among nurses need to be sought.