• The Korean Journal of Applied Statistics
• /
• v.19 no.2
• /
• pp.305-317
• /
• 2006

Outliers adversely affect model identification, parameter estimation, and forecast in time series data. In particular, when outliers consist of a patch of additive outliers, the current outlier detection procedures suffer from the masking and swamping effects which make them inefficient. In this paper, we propose new outlier detection procedure based on high breakdown estimators, called as the dual robust filtering. Empirical and simulation studies in the autoregressive model with orders p show that the proposed procedure is effective.

• Explosives and Blasting
• /
• v.23 no.4
• /
• pp.31-44
• /
• 2005

In highly stressed conditions, the excavation damage zone induced by stress redistribution and disturbance must be evaluated after tunnel excavation. Therefore, the investigation of stress-induced deformation and fracture in rock is indispensable. In this study, fracture and damage mechanisms of rock induced by the accumulation of microcracks were investigated by the moving point regression technique as well as acoustic emission measured during uniaxial compression tests. Especially, the modified procedures to determine damage thresholds more systematically were newly proposed, and successfully applied to rock. From experiments, crack initiation and track damage stress levels were estimated to be $33{\~}36\%$ and $84{\~}89\%$ of uniaxial compressive strength respectively, for both of Hwangdeung granite and Yeosan marble. However, the normalized crack closure stress level for Yeosan marble was much higher than for Hwangdeung granite. In addition, the largest proportion of total axial strain in Hwangdeung granite was attributable to elastic deformation and initial microcracking. However, the greatest part of axial deformation in Yeosan marble arose from initial crack closure and unstable cracking. Finally, it was seen that unstable cracking after the crack damage stress level played a key part in the lateral deformation in rocks under uniaxial compression.

• Journal of the Korean Wood Science and Technology
• /
• v.38 no.6
• /
• pp.526-531
• /
• 2010

For efficient use of wood, it is important to control moisture of wood in processing wood. Near-infrared (NIR) spectroscopy can be used to estimate the physical and chemical properties of materials quickly and nondestructively. In this study, it was intended to measure the moisture contents on the surface of wood using NIR spectroscopy coupled with multivariate analytic statistical techniques. Because NIR spectroscopy is affected by the chemical components of the specimens and contains signal noise, a regression model for detecting moisture content of wood was established after carrying out several numerical pretreatments such as Smoothing, Derivative and Normalization in this study. It shows that the regression model using NIR absorbance in the range of 750~2,500 nm predicts the actual surface moisture content very well. Near-infrared spectroscopy technique developed in this study is expected to improve a technology to control moisture content of wood in using and drying process.

• Analytical Science and Technology
• /
• v.25 no.4
• /
• pp.250-256
• /
• 2012

Two marker compounds of Achyranthis Radix, ecdysterone and inokosterone, were analyzed by HPLC on an ODS column ($250{\times}4.6$ mm, 5 ${\mu}m$) with a mobile phase of 15% acetonitrile containing 0.08% formic acid at a flow rate of 1.0 mL/min and a detection wavelength of UV 254 nm. The method was validated by ICH guideline and applied to the monitoring of marker compounds in 93 samples of Achyranthis Radix collected at various areas in Korea and China. The new content criteria of ecdysterone and inokosterone, established using linear regression method were 0.033% and 0.020%, respectively. When the new content criteria were applied to the quality control test of commercial Achyranthis Radix, 95.4% of total samples including 100% of Korean and 92.6% of Chinese samples were passed the test. Application of new content criteria could protect the Korean products and decrease the distribution of Chinese products with lower quality.

• Korean Journal of Food Science and Technology
• /
• v.28 no.2
• /
• pp.345-351
• /
• 1996

Studies on extraction and color characteristics of purple sweet potato (PSP) pigment were performed to provide the basic information for the utilization of PSP as a new source of natural food colorant. PSP pigment was extracted well with the polar solvents such as distilled water, ethanol, and methanol. but hardly extracted with the non-polar solvents. Among the tested solvents, 20% ethanol solution containing 0.1% citric acid was found to be the most efficient for extraction of the pigment from PSP. PSP contained high amount of pigment not only in the epidermis but also in the flesh of the potato. The PSP pigment was heat stable even under pretreatments such as autoclaving and blanching of the potato before extraction. The optimum temperature of the extraction for the PSP Pigment was decided to be $30^{\circ}C$ by considering the stability and the rate of extraction. The pigment was markedly influenced by the change of pH. The color of the pigment solution was red at the pH range of $1.0{\sim}3.0$, became blue at $7.0{\sim}8.0$, then turned green at $9.0{\sim}10.0$. A characteristic batho-chromic shift of the pigment solution was observed as the pH of the solution increased.

• Journal of Intelligence and Information Systems
• /
• v.27 no.1
• /
• pp.23-46
• /
• 2021

Collaborative filtering, which is often used in personalization recommendations, is recognized as a very useful technique to find similar customers and recommend products to them based on their purchase history. However, the traditional collaborative filtering technique has raised the question of having difficulty calculating the similarity for new customers or products due to the method of calculating similaritiesbased on direct connections and common features among customers. For this reason, a hybrid technique was designed to use content-based filtering techniques together. On the one hand, efforts have been made to solve these problems by applying the structural characteristics of social networks. This applies a method of indirectly calculating similarities through their similar customers placed between them. This means creating a customer's network based on purchasing data and calculating the similarity between the two based on the features of the network that indirectly connects the two customers within this network. Such similarity can be used as a measure to predict whether the target customer accepts recommendations. The centrality metrics of networks can be utilized for the calculation of these similarities. Different centrality metrics have important implications in that they may have different effects on recommended performance. In this study, furthermore, the effect of these centrality metrics on the performance of recommendation may vary depending on recommender algorithms. In addition, recommendation techniques using network analysis can be expected to contribute to increasing recommendation performance even if they apply not only to new customers or products but also to entire customers or products. By considering a customer's purchase of an item as a link generated between the customer and the item on the network, the prediction of user acceptance of recommendation is solved as a prediction of whether a new link will be created between them. As the classification models fit the purpose of solving the binary problem of whether the link is engaged or not, decision tree, k-nearest neighbors (KNN), logistic regression, artificial neural network, and support vector machine (SVM) are selected in the research. The data for performance evaluation used order data collected from an online shopping mall over four years and two months. Among them, the previous three years and eight months constitute social networks composed of and the experiment was conducted by organizing the data collected into the social network. The next four months' records were used to train and evaluate recommender models. Experiments with the centrality metrics applied to each model show that the recommendation acceptance rates of the centrality metrics are different for each algorithm at a meaningful level. In this work, we analyzed only four commonly used centrality metrics: degree centrality, betweenness centrality, closeness centrality, and eigenvector centrality. Eigenvector centrality records the lowest performance in all models except support vector machines. Closeness centrality and betweenness centrality show similar performance across all models. Degree centrality ranking moderate across overall models while betweenness centrality always ranking higher than degree centrality. Finally, closeness centrality is characterized by distinct differences in performance according to the model. It ranks first in logistic regression, artificial neural network, and decision tree withnumerically high performance. However, it only records very low rankings in support vector machine and K-neighborhood with low-performance levels. As the experiment results reveal, in a classification model, network centrality metrics over a subnetwork that connects the two nodes can effectively predict the connectivity between two nodes in a social network. Furthermore, each metric has a different performance depending on the classification model type. This result implies that choosing appropriate metrics for each algorithm can lead to achieving higher recommendation performance. In general, betweenness centrality can guarantee a high level of performance in any model. It would be possible to consider the introduction of proximity centrality to obtain higher performance for certain models.