• The Korean Journal of Applied Statistics
• /
• v.33 no.2
• /
• pp.115-122
• /
• 2020

K-means clustering uses a spherical or elliptical metric to group data points; however, it does not work well for non-convex data such as the concentric circles. Spectral clustering, based on graph theory, is a generalized and robust technique to deal with non-standard type of data such as non-convex data. Results obtained by spectral clustering often outperform traditional clustering such as K-means. In this paper, we review spectral clustering and show important issues in spectral clustering such as determining the number of clusters K, estimation of scale parameter in the adjacency of two points, and the dimension reduction technique in clustering high-dimensional data.

• The Korean Journal of Applied Statistics
• /
• v.15 no.2
• /
• pp.269-279
• /
• 2002

We considered the problem for confirmation of multiple outliers and leverage points. Identification of multiple outliers and leverage points is difficult because of the masking effect and swamping effect. Rousseeuw and van Zomeren(1990) identified multiple outliers and leverage points by using the Least Median of Squares and Minimum Value of Ellipsoids which are high-breakdown robust estimators. But their methods tend to declare too many observations as extremes. Atkinson(1987) suggested a method for confirming of outliers and Fung(1993) pointed out Atkinson method's limitation and proposed another method by using the add-back model. But we analyzed that Fung's method is affected by adjacent effect. In this thesis, we proposed one procedure for confirmation of outliers and leverage points and compared three example with Fung's method.

• Journal of Korea Technology Innovation Society
• /
• v.6 no.3
• /
• pp.279-297
• /
• 2003

The purpose of this paper is to analyze the determinants of company R＆D investment with zero observations by using the data of R＆D Scoreboard published by Ministry of Science and Technology(2002). Conventional parametric approach to dealing with zero investments is not robust to heteroscedastic and/or non-normal error structure. Thus, this study applies symmetrically trimmed least squares(STLS) estimation as a semi-parametric approach to dealing with zero R&D investments. The result of specification test indicates the semi-parametric approach outperforms the parametric approach significantly. Moreover, the results of the study provide various implications as summarized below. The R&D investment of IT company is larger than that of non-IT company. The R&D investment has a positive relation to foreigners' investment ratio. The higher degree of financial self-reliance is, the larger the R&D investment is. Firm size variables such as sales amount and the number of workers are positively related to R&D investment. The sales elasticity of R&D investment is larger than one. However, the workers elasticity of R&D investment is smaller than one.

• The Korean Journal of Applied Statistics
• /
• v.31 no.2
• /
• pp.229-239
• /
• 2018

This paper introduces how to adopt the concept of temporal hierarchies to forecast time series data. Similarly as in hierarchical cross-sectional data, temporal hierarchies can be constructed for any time series data by means of non-overlapping temporal aggregation. Reconciliation forecasts with temporal hierarchies result in more accurate and robust forecasts when compared with the independent base and bottom-up forecasts. As an empirical example, we forecast traffic accident counts with temporal hierarchies and observe that reconciliation forecasts are superior to the base and bottom-up forecasts in terms of forecast accuracy.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.13 no.1
• /
• pp.91-96
• /
• 2003

The forces and moments at the aircraft c.g. have components due to aerodynamic effects and to engine thrust. For the flight stability and autopilot systems we present a attitude control method using an intelligent control algorithm Which is based on the control rules from experts knowledge concerning the motion equations and other experiences. Then a robust fuzzy controller is developed to control the flight attitude. The controller can deal with multiple inputs and outputs. We have made an aircraft model and the orientation sensor for experimental flights. The control rules based on the flight expert s experience and knowledge can be programmed by fuzzy rules, and determined control rules by experimental flight. We can be stable attitude control by fuzzy controller.

• Journal of Korean Society for Quality Management
• /
• v.39 no.4
• /
• pp.507-515
• /
• 2011

MIL-HDBK-217 has played a pivotal role in reliability prediction of electronic equipments for more than 30 years. Recently, RIAC developed a new methodology $217Plus^{TM}$which officially replaces MIL-HDBK-217. Sensitivity analysis of the 217Plus component models to various parameters has been performed and meaningful observations have been drawn in this study. We first briefly reviewed the $217Plus^{TM}$ methodolog and compared it with the conventional model, MIL-HDBK-217. We then performed sensitivity analysis $217Plus^{TM}$ component models to various parameters. Based on the six parameters and an orthogonal array selected, we have performed indepth analyses concerning parameter effects on the model. Our result indicates that, among various parameters, operating temperature and temperature rise during operation have the most significant impacts on the life of a component, and thus a design robust to high temperature is the most importantly required. Next, year of manufacture, duty cycle, and voltage stress are weaker but may be significant when they are in heavy load conditions. Although our study is restricted to a specific type of diodes, the results are still valid to other cases. The results in this study not only figure out the behavior of the predicted failure rate as a function of parameters but provide meaningful guidelines for practical applications.

• Transactions of the Korean Society of Mechanical Engineers B
• /
• v.22 no.11
• /
• pp.1591-1602
• /
• 1998

Simple spanwise distribution models of deviation angle and pressure loss coefficient due to the tip leakage flow are formulated for use in association with the streamline curvature method as a flow analysis. Combining these new models with the previous deviation and loss models due to secondary flow, a robust streamline curvature method is established for flow analysis of single-stage, subsonic axial turbines with wide ranges of turning angle, aspect ratio and blading type. At the exit from rotor rows, the flow variables are mixed radially according to a spanwise transport equation. The proposed streamline curvature method is tested against a forced vortex type turbine as well as a free vortex type one. The results show that the spanwise variations of flow angle, axial velocity and loss coefficients at rotor exit are predicted with good accuracy, being comparable to a steady three-dimensional Navier-Stokes analysis. This simple and fast flow analysis is found to be very useful for the turbine design at the initial design phase.

• Journal of the Korean Society for Library and Information Science
• /
• v.36 no.3
• /
• pp.259-274
• /
• 2002

Recently, a great deal of corporations have adopted knowledge management system with eagerness to enhance the company competitiveness. However, since the main feature of knowledge management system is not just a simple information system but another entity creating intangible assets tailed "knowledge", we need to develop a new approach to investigating the performance and acceptance of knowledge management systems from a perspective allowing knowledge-sensitive constructs. In this regard, we develop new constructs like knowledge schema and several knowledge circulation-related activities. As a research model, we adopt a famous technology acceptance model or TAM by Davis (1989), and extend it into incorporating knowledge schema. With the statistically valid and usable questionnaire survey data collected from 886 respondents in a big corporation typically using knowledge management system, we induced a robust result empirically, saying that knowledge schema and knowledge circulation activities are valid determinants of performance and acceptance of knowledge management systems.

• Proceedings of the Korean Society of Propulsion Engineers Conference
• /
• 1998.04a
• /
• pp.13-13
• /
• 1998

A simple model of the tip leakage flow models of the rotor downstream flow is developed, based on Lakshminarayana's theoretical concept on the tip clearance flow and the experimental data published in open literature. And new spanwise distribution models of deviation angle and pressure loss coefficient due to the tip leakage flow are formulated for use in association with the streamline curvature method as a through flow analysis. Combining these new models and previous deviation and loss models due to secondary flow, a robust streamline curvature method is established for flow analysis of single-stage, subsonic axial turbines with wide ranges of turning angle, aspect ratio and blading type. At the exit from rotor rows, the flow variables are mixed radially according to a spanwise transport equation. The proposed streamline curvature method is tested against a forced vortex type turbine as well as a free vortex type one. The results show that the spanwise variations of flow angle, axial velocity and loss coefficients at rotor exit are predicted with good accuracy, being comparable to a steady three-dimensional Navier-Stokes analysis. This simple and fast flow analysis is found to be very useful for the turbine design at the initial design phase.

• The Korean Journal of Applied Statistics
• /
• v.5 no.2
• /
• pp.255-268
• /
• 1992

We propose an iterated interpolation approach for the estimation fo time series parameters in the presence of outliers. The proposed approach iterates the parameter estimation stage and the outlier detection stage until no further outliers are detected. For the detection of outliers, interpolation diagnostic is applied, where the atypical observations by the one-step-ahead predictor instead of downweighting is also proposed. The performance of the proposed estimation methods is compared with other robust estimation methods by simulation study. It is observed that the iterated interpolation approach performs reasonably well is general, especially for single AO case and large $\phi$ in absolute values.