• The Korean Journal of Applied Statistics
• /
• v.22 no.3
• /
• pp.531-539
• /
• 2009

Logistic regression is widely used as a datamining technique for the customer relationship management. The maximum likelihood estimator has highly inflated variance when multicollinearity exists among the regressors, and it is not robust against outliers. Thus we propose the robust principal components logistic regression to deal with both multicollinearity and outlier problem. A procedure is suggested for the selection of principal components, which is based on the condition index. When a condition index is larger than the cutoff value obtained from the model constructed on the basis of the conjoint analysis, the corresponding principal component is removed from the logistic model. In addition, we employ an algorithm for the robust estimation, which strives to dampen the effect of outliers by applying the appropriate weights and factors to the leverage points and vertical outliers identified by the V-mask type criterion. The Monte Carlo simulation results indicate that the proposed procedure yields higher rate of correct classification than the existing method.

• Proceedings of the Korean Association of Geographic Inforamtion Studies Conference
• /
• 2003.04a
• /
• pp.151-156
• /
• 2003

본 연구에서는 2000년 LANDSAT ETM+ 수도권 영상을 이용하여 도시지역 10개소, 식생지역 10개소를 선정해서 각각에 대해 표준화 주성분 분석을 적용하여 두 지역간의 고유벡터 매트릭스를 비교ㆍ분석해보았다. 도시 지역과 식생 지역각각에 대해 총 6개의 주성분이 생성되었으며 PC-2와 고유벡터 부호가 변한 밴드(band2, band7)를 RGB로 조합하여 수원지역을 대상으로 분류(Classification)한 결과의 정확도를 분광서명 분별 분석(Signature Separability Analysis)통해 얻은 밴드조합(band1, band3, band5) 영상의 분류결과와 비교해 보았다. 수원지역 2000년 IKONOS 영상의 다중분광 밴드(4×4m)와 전정색 밴드(1x1m)를 융합한 영상이 분류 정확도를 판단하는 기준으로 사용되었다. 비교결과 분류 전체 정확도는 각각 87.7%, 77.29% Khat 지수는 0.83, 0.68로 나타나 PC-2, 밴드2, 밴드7을 이용했을 때 분류 정확도를 높일 수 있다는 결과를 얻었다.

• Journal of Bio-Environment Control
• /
• v.17 no.2
• /
• pp.124-131
• /
• 2008

To investigate main air temperature factors correlated to difference of fruit characteristics according to cultivating areas, fruit and air temperature characteristics of eight cultivating areas of 'Fuyu' persimmon were analyzed by principle components and multiple regression analysis. The first principal components extracted from 16 air temperature factors was annual mean temperature, mean temperature during October, annual mean minimum extreme temperature, mean temperature during growing period, and so forth. The second principal components was mean temperature during May and June and so forth. And cumulative contribution was 91.4%. The five of eight cultivating area had clearly the difference of main factors or the correlated direction among cultivating areas. In multiple regression analysis between the extracted main factors and fruit characteristics, fruit hight were highly correlated with mean temperature during growing period ($X_8$) and cumulative temperature ($X_6$), and the regression equation was $Y=150.55-5.375X_8+ 0.014X_6(r^2=0.843)$. Also this regression equation was affected by mean minimum temperature during growing period, cumulative temperature, and mean temperature during August. Fruit diameter was negatively correlated with mean temperature during growing period, flesh browning rate and Hunter a value of peel color were positively correlated with mean minimum temperature during growing period and annual minimum air temperature, respectively.

• Proceedings of the Korea Multimedia Society Conference
• /
• 2003.05b
• /
• pp.435-440
• /
• 2003

본 논문에서는 실시간 출입통제시스템에 적용이 가긍한 복잡한 배경에서의 다중 얼굴 영역 검출과 추적을 통한 얼굴 인식 알고리즘을 제안하였다. 제안된 알고리즘에서는 배경영상과 입력된 연속적인 프레임간의 차영상을 적용함으로써 물체의 움직임을 감지한 후. IISI컬러 좌표모델을 이용하여 얼굴의 1차 후보 영역을 검출하고, 잡음제거를 위해 모폴로지 연산을 수행하였다 또한 Line Projection을 이용한 객체 분할법(Object Segmentation)으로 객체를 분할함으로써 다중 얼굴 영역을 추출하였다. 또한 추출된 얼굴영역에서 눈 영역 검출을 통해 각각의 얼굴 영역들을 검증하였으며 검증된 얼굴들의 최외각 4개의 좌표를 이용하여 얼굴 추적율을 높였다. 마지막으로 얼굴 인식은 추출된 얼굴 영역으로부터 주성분 분석(PCA ： Principle Component Analysis)방법을 이용함으로써 97～98％의 높은 인식율을 보였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2020.05a
• /
• pp.397-400
• /
• 2020

본 논문은 영화에 나오는 효과음을 자막으로 생성해주는 자동자막생성을 제안하며, 그의 첫 단계로써 다중 사운드 분류 모델을 제안하였다. 고양이, 강아지, 사람의 음성을 분류하기 위해 사운드 데이터의 특정벡터를 추출한 뒤, 4가지의 기계학습에 적용한 결과 최적모델로 딥러닝이 선정되었다. 전처리 과정 중 주성분 분석의 유무에 따라 정확도는 81.3%와 33.3%로 확연한 차이가 있었으며, 이는 복잡한 특징을 가지는 사운드를 분류하는데 있어 주성분 분석과 넓고 깊은 형태의 신경망이 보다 개선된 분류성과를 가져온 것으로 생각된다.

• The Journal of Bigdata
• /
• v.5 no.2
• /
• pp.135-143
• /
• 2020

This paper proposes a pre-processing method and a dimensional reduction method in the analysis of shopping carts where there are many correlations between variables when dividing the types of consumers in the agri-food consumer panel data. Cluster analysis is a widely used method for dividing observational objects into several clusters in multivariate data. However, cluster analysis through dimensional reduction may be more effective when several variables are related. In this paper, the food consumption data surveyed of 1,987 households was clustered using the K-means method, and 17 variables were re-selected to divide it into the clusters. Principal component analysis and factor analysis were compared as the solution for multicollinearity problems and as the way to reduce dimensions for clustering. In this study, both principal component analysis and factor analysis reduced the dataset into two dimensions. Although the principal component analysis divided the dataset into three clusters, it did not seem that the difference among the characteristics of the cluster appeared well. However, the characteristics of the clusters in the consumption pattern were well distinguished under the factor analysis method.

• Journal of the Korean Society for Railway
• /
• v.17 no.5
• /
• pp.381-389
• /
• 2014

This research focuses on the current trend of 'Sharing Transportation' to clarify the regional factors having a decisive effect on the use of Car Sharing. To accomplish this, the current research is built a Database of the regional characteristics of Car Sharing spots based on railway stations in Seoul and performed an analysis of the primary regional factors affecting Car Sharing usage. As a result, we found conclusive factors affecting the use of Car Sharing. This research can be utilized for establishing strategies and effective measures to support the use of Car Sharing and sustainable development with respect to issues of motorization.

• Journal of Navigation and Port Research
• /
• v.30 no.1 s.107
• /
• pp.35-43
• /
• 2006

This study suggests strategies to prevent traffic accidents by utilizing impact factors per each cluster and the typical patterns of 81 cities based on the statistical analysis of the data concerning the TCI which was developed from the partnership of the Traffic Safety Authority and the Green Traffic Movement Corporation in 2002 and 2003. The Principal Component Analysis and Cluster Analysis on impact factors and TCI result in 4 components and 4 clusters. Also as the results of Stepwise Multiple Regression Analysis examining the relationship between impact factors and TCI, R2 values of these models show high to all clusters. According to the results, we suggest strategies to prevent traffic accidents per cluster concretely and it is necessary to analyze how effective the invested facilities are in reducing traffic accidents in the future.

• Journal of the Korean Society for Nondestructive Testing
• /
• v.23 no.3
• /
• pp.254-262
• /
• 2003

In this study, acoustic emission (AE) signals due to surface cracking and moisture movement in the flat-sawn boards of oak (Quercus Variablilis) during drying under the ambient conditions were analyzed and classified using the principal component analysis. The AE signals corresponding to surface cracking showed higher in peak amplitude and peak frequency, and shorter in rise time than those corresponding to moisture movement. To reduce the multicollinearity among AE features and to extract the significant AE parameters, correlation analysis was performed. Over 99% of the variance of AE parameters could be accounted for by the first to the fourth principal components. The classification feasibility and success rate were investigated in terms of two statistical classifiers having six independent variables (AE parameters) and six principal components. As a result, the statistical classifier having AE parameters showed the success rate of 70.0%. The statistical classifier having principal components showed the success rate of 87.5% which was considerably than that of the statistical classifier having AE parameters.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2017.05a
• /
• pp.154-154
• /
• 2017

전 세계적인 기후변동과 기후변화의 영향으로 대규모 인명 및 재산피해를 유발하는 자연재난의 빈도와 강도가 증가하고 있다. 이렇게 변화하는 상황에서 효율적인 대책을 수립하기 위해서는 재해에 노출된 특성을 지역적 특성과 함께 고려하여 지역별로 재해에 위험한 정도를 평가하는 것이 선행되어지고, 재난 피해 발생전에 피해 지역 및 범위를 예측하는 것이 필요하다고 판단된다. 따라서 본 연구에서는 국내 자연재난 피해의 65% 이상을 차지하는 호우피해를 대상으로 PSR(Pressure-State-Response) 구조를 이용하여 호우피해위험지수(Heavy rain Damage Risk Index, HDRI)를 제안하여 호우 위험도를 평가하고자하였다. 또한 도출된 지역별 위험등급에 따른 호우피해 예측함수를 개발하여 재해발생 전에 개략적인 피해의 범위를 예측하고자 하였다. 먼저 지역별 호우 위험도 평가를 위해 압력지표, 현상지표, 대책지표를 구축하고, 주성분분석을 이용하여 평가지표를 결정하였다. 결정된 평가지표를 동일한 가중치를 부여하여 호우피해위험지수를 도출하였다. 분석결과, 경기도 31개 지자체 중에서 가장 안전한 1등급인 지자체는 15개의 지자체로 나타났으며, 2등급인 지자체는 7개, 3등급인 지자체는 9개로 분류되었다. 지자체별 호우 위험도 등급에 따라서 재해기간별 총강우량, 재해일수, 선행강우량(1~5일), 지속시간별 최대강우량(1~24시간) 등의 자료를 설명변수로 구축하였고, 다중회귀모형과 주성분분석을 활용하여 예측함수를 개발하였다. 등급별 호우피해 예측함수는 N-RMSE가 12~18%로 호우피해를 적절하게 예측하는 것으로 평가되었다. 본 연구를 통해 지자체별 호우피해위험도 등급을 파악 할 수 있으며, 평가된 호우피해위험도 등급별로 호우피해 예측함수 개발을 통해 사전에 호우피해 발생 및 규모를 파악할 수 있게 되었다. 따라서 본 연구의 결과는 각 지자체 및 관련 부처에서 효과적인 방재체계를 수립하는데 있어 기초자료로 활용될 수 있을 것으로 판단된다.