• 한국정밀공학회지
• /
• 제16권12호
• /
• pp.119-125
• /
• 1999

This paper gives an overview of the various approaches to designing statistical pattern recognition scheme based on Bayes discrimination rule and the artificial neural networks for rotating machine condition classification. Concerning to Bayes discrimination rule, this paper contains the linear discrimination rule applied to classification into several multivariate normal distributions with common covariance matrices, the quadratic discrimination rule under different covariance matrices. Also we discribes k-nearest neighbor method to directly estimate a posterior probability of each class. Five features are extracted in time domain vibration signals. Employing these five features, statistical pattern classifier and neural networks have been established to detect defects on rotating machine. Four different cases of rotation machine were observed. The effects of k number and neural networks structures on monitoring performance have also been investigated. For the comparison of diagnosis performance of these two method, their recognition success rates are calculated form the test data. The result of experiment which classifies the rotating machine conditions using each method presents that the neural networks shows the highest recognition rate.

• Journal of Microbiology and Biotechnology
• /
• 제23권7호
• /
• pp.932-941
• /
• 2013

This work aimed to classify Aspergillus (8 species, 28 strains) by using a secondary metabolite profile-based chemotaxonomic classification technique. Secondary metabolites were analyzed by liquid chromatography ion-trap mass spectrometry (LC-IT-MS) and multivariate statistical analysis. Most strains were generally well separated from each section. A. lentulus was discriminated from the other seven species (A. fumigatus, A. fennelliae, A. niger, A. kawachii, A. flavus, A. oryzae, and A. sojae) with partial least-squares discriminate analysis (PLS-DA) with five discriminate metabolites, including 4,6-dihydroxymellein, fumigatin, 5,8-dihydroxy-9-octadecenoic acid, cyclopiazonic acid, and neosartorin. Among them, neosartorin was identified as an A. lentulus-specific compound that showed anticancer activity, as well as antibacterial effects on Staphylococcus epidermidis. This study showed that metabolite-based chemotaxonomic classification is an effective tool for the classification of Aspergillus spp. with species-specific activity.

• 응용통계연구
• /
• 제26권2호
• /
• pp.223-236
• /
• 2013

미국 프로야구(Major League Baseball) 명예의 전당의 입성과 탈락을 예측할 수 있는 여러 가지 통계적인 분류분석법을 실시하고 그 결과의 정확성을 비교하였다. 이를 위해 명예의 전당 가입 조건을 만족하는 타자들 중 1980년 이후 기록된 데이터의 17개의 독립변수를 사용하여 분류분석에서 널리 사용되는 기준으로 판별분석, 로지스틱 회귀분석과 상대적으로 최근에 제안된 Mahalanobis-Taguchi System(MTS)을 실시하여 비교하였다. 이 세 가지 방법 중 MTS가 상대적으로 더 나은 효율을 보였으며 이는 다변량 관측 값이 방향성이 없어 속성에 따른 도형적 그룹을 형성하지 못하는 경우에 효율적인 MTS의 특성에 의한 것으로 판단된다.

• 한국컴퓨터정보학회논문지
• /
• 제25권6호
• /
• pp.35-47
• /
• 2020

박스 오피스 예측은 영화 이해관계자들에게 중요하다. 따라서 정확한 박스 오피스 예측과 이에 영향을 미치는 주요 변수를 선별하는 것이 필요하다. 본 논문은 영화의 박스 오피스 예측 정확도 향상을 위해 다변량 시계열 데이터 분류와 주요 변수 선택 방법을 제안한다. 연구 방법으로 한국 영화 일별 데이터를 KOBIS와 NAVER에서 수집하였고, 랜덤 포레스트(Random Forest) 방법으로 주요 변수를 선별하였으며, 딥러닝(Deep Learning)으로 다변량 시계열을 예측하였다. 한국의 스크린 쿼터제(Screen Quota) 기준, 딥러닝을 이용하여 영화 개봉 73일째 흥행 예측 정확도를 주요 변수와 전체 변수로 비교하고 통계적으로 유의한지 검정하였다. 딥러닝 모델은 다층 퍼셉트론(Multi-Layer Perceptron), 완전 합성곱 신경망(Fully Convolutional Neural Networks), 잔차 네트워크(Residual Network)로 실험하였다. 결과적으로 주요 변수를 잔차 네트워크에 사용했을 때 예측 정확도가 약 93%로 가장 높았다.

• 응용통계연구
• /
• 제22권5호
• /
• pp.1097-1102
• /
• 2009

Classification of subjects with unknown distribution in small sample size setup may involve order-restricted constraints in multivariate parameter setups. Those problems makes optimality of conventional likelihood ratio based statistical inferences not feasible. Fortunately, Roy (1953) introduced union-intersection principle(UIP) which provides an alternative avenue. Redescending M-estimator along with that principle yields a considerably appropriate robust testing procedure. Furthermore, conditionally distribution-free test based upon exact permutation theory is used to generate p-values, even in small sample. Applications of this method are illustrated in simulated data and read data example (Lobenhofer et al., 2002)

• Communications for Statistical Applications and Methods
• /
• 제16권5호
• /
• pp.881-889
• /
• 2009

Support vector machines(SVMs) are known as flexible and efficient classifier of multivariate observations, producing a hyperplane or hyperdimensional curved surface in multidimensional feature space that best separates training samples by known groups. As various methodological extensions are made for SVM classifiers in recent years, it becomes more difficult to understand the constructed model intuitively. The aim of this paper is to visualize various SVM classifications tuned by several parameters in reduced dimensions, so that data analysts secure the tangible image of the products that the machine made.

• Journal of the Korean Data and Information Science Society
• /
• 제5권1호
• /
• pp.1-10
• /
• 1994

Projection pursuit(PP) is a computer-intensive method which seeks out interesting linear projections of multivariate data onto a lower dimension space by machine. By working with lower dimensional projections, projection pursuit avoids the sparseness of high dimensional data. We show through simulation that two projection pursuit discriminant mothods proposed by Chen(1989) and Huber(1985) do not improve very much the error rate than the existing methods and compare several classification procedures.

• Archives of Plastic Surgery
• /
• 제40권5호
• /
• pp.522-529
• /
• 2013

Background Despite advances in surgical techniques, sterile protocols, and perioperative antibiotic regimens, surgical site infections (SSIs) remain a significant problem. We investigated the relationship between wound classification (i.e., clean, clean/contaminated, contaminated, dirty) and SSI rates in plastic surgery. Methods We performed a retrospective review of a multi-institutional, surgical outcomes database for all patients undergoing plastic surgery procedures from 2006-2010. Patient demographics, wound classification, and 30-day outcomes were recorded and analyzed by multivariate logistic regression. Results A total of 15,289 plastic surgery cases were analyzed. The overall SSI rate was 3.00%, with superficial SSIs occurring at comparable rates across wound classes. There were similar rates of deep SSIs in the clean and clean/contaminated groups (0.64%), while rates reached over 2% in contaminated and dirty cases. Organ/space SSIs occurred in less than 1% of each wound classification. Contaminated and dirty cases were at an increased risk for deep SSIs (odds ratios, 2.81 and 2.74, respectively); however, wound classification did not appear to be a significant predictor of superficial or organ/space SSIs. Clean/contaminated, contaminated, and dirty cases were at increased risk for a postoperative complication, and contaminated and dirty cases also had higher odds of reoperation and 30-day mortality. Conclusions Analyzing a multi-center database, we found that wound classification was a significant predictor of overall complications, reoperation, and mortality, but not an adequate predictor of surgical site infections. When comparing infections for a given wound classification, plastic surgery had lower overall rates than the surgical population at large.

• 대한원격탐사학회지
• /
• 제15권3호
• /
• pp.195-208
• /
• 1999

위성영상 분류작업에서 분류클래스에 대한 샘플화소의 대표성은 분류 정확도에 많은 영향을 미친다. 따라서, 통계적 영상분류방법에서는 분류 기법 자체보다 분류 확률을 결정하는 트레이닝 단계, 즉 샘플화소의 최적화가 필요하다. 본 연구에서는 SPOT XS, LANDSAT TM을 이용한 위성영상 화소분류작업에서 분류 이전단계, 즉 샘플화소의 정규성을 계산하여, 정규성에 악영향을 미치는 화소를 객관적 기준으로 조정하였다. 정규화과정을 위한 유전자 알고리즘 적용의 생존확률 평가함수로 다변량 Q-Q plot의 상관계수와 트레이닝의 분산값을 고려하였으며, 5% 유의수준을 적용하였다. 연구결과, 실험대상지역의 경우, 유전자 알고리즘을 이용한 트레이닝 정규화 결과가 대부분의 클래스에 대하여 그 평균과 분산을 모집단에 근사시키고 있다는 것을 입증하였고, 해당 클래스의 모집단 분포를 예측할 수 있는 가능성을 제시하였다.

• 한국의류학회지
• /
• 제23권4호
• /
• pp.563-574
• /
• 1999

The purpose of this study was to provide basic information for obese boy's clothing construction that can reflect the characteristics of their bodies. The subjects for anthropometric measurements which were performed directly were obese boys of 9 to 11 year-old. To classify the somatotype and to analyze the characteristics of each somatotype 310 obese boys were examined. Data were analyzed by using multivariate method, By means of Ward the subjects were classified into 4 clusters according to the factor scores which were obtained from 6 factors providing the information of 54 items. 4 clusters were identified. 1) Type I was characterized by tall and obese type 2) Type II was characterized by short and small type 3) Type III was characterized by long and obese type of lower body. 4) Type IV was characterized by short and obese type.