• Communications for Statistical Applications and Methods
• /
• v.7 no.3
• /
• pp.759-766
• /
• 2000

In order to design fractional factorial experiments which include some debarred combinations, we should select defining contrasts so that those combinations are to be excluded. Choi(1999) presented a method of selectign defining contrasts to construct orthogonal 3-level fractional factorial experiments which exclude some debarred combinations. In this paper, we extend Choi's method to general p-level fractional factorial experiments to select defining contrasts which cold exclude some debarred combinations.

• Journal of Korean Society for Quality Management
• /
• v.49 no.1
• /
• pp.31-45
• /
• 2021

Purpose: A noncentral composite design method is to be developed to explore farther region for the first factorial design. A general guideline for sequential experimentation is provided. Methods: (1) A non-overlapping noncentral composite design (NNCD) is developed, in which the second factorial design shares one design point that indicates the best response value in the first factorial design. (2) Four composite designs are compared in terms of the four design evaluation criteria, which are D-, A, G, and I-optimality. (3) A follow-up design strategy is suggested based on the interaction effect, direction of improvement, number of factors. Results: (1) NNCD and model building method are presented, which is useful for exploring farther region from first factorial design block. (2) The performances of the four composite designs are compared. (3) A follow-up design strategy is suggested. Conclusion: (1) NNCD will be useful to explore farther region for the first factorial design. (2) A follow-up design strategy can be beneficial to the experimental practitioners for product and process design and improvement.

• Communications for Statistical Applications and Methods
• /
• v.14 no.3
• /
• pp.483-495
• /
• 2007

In a general fractional factorial design, the n-levels of a factor are coded by the $n^{th}$ roots of the unity. Pistone and Rogantin (2007) gave a full generalization to mixed-level designs of the theory of the polynomial indicator function using this device. This article discusses the optimal blocking scheme for nonregular designs. According to hierarchical principle, the minimum aberration (MA) has been used as an important criterion for selecting blocked regular fractional factorial designs. MA criterion is mainly based on the defining contrast groups, which only exist for regular designs but not for nonregular designs. Recently, Cheng et al. (2004) adapted the generalized (G)-MA criterion discussed by Tang and Deng (1999) in studying $2^p$ optimal blocking scheme for nonregular factorial designs. The approach is based on the method of replacement by assigning $2^p$ blocks the distinct level combinations in the column with different blocks. However, when blocking level is not a power of two, we have no clue yet in any sense. As an example, suppose we experiment during 3 days for 12-run Plackett-Burman design. How can we arrange the 12-runs into the three blocks? To solve the problem, we apply G-MA criterion to nonregular mixed-level blocked scheme via the mixed-level indicator function and give an answer for the question.

• Journal of the Korean Statistical Society
• /
• v.10
• /
• pp.122-127
• /
• 1981

A parallel flats fraction for the $3^n$ factorial is symbolically written as $At=C=(C_1 C_2 \cdots C_f)$ where C is a rxf matrix and A is rxn matrix with rank r. It is shown that the set of all possible parallel flats fraction C for a given A and given size can be partitioned into equivalence classes. The number of those classes are enumerated in general.

• 한국데이터정보과학회:학술대회논문집
• /
• 2006.04a
• /
• pp.121-129
• /
• 2006

Two color or cDNA microarrays are extensively used to study relative expression levels of thousands of genes simultaneously. 0かy two tissue samples can be hybridized on a single microarray slide. Thus, a microarray slide necessarily forms an incomplete block design with block size two when more than two tissue samples are under study. We also need to control for variability in gene expression values due to the two dyes. Thus, red and green dyes form the second blocking factor in addition to slides. General design problem for these microarray experiments is discussed in this paper. Designs for factorial cDNA microarrays are also discussed.

• Journal of the Korean Data and Information Science Society
• /
• v.22 no.2
• /
• pp.245-253
• /
• 2011

Powers of rank transformed statistic for testing main effects and interaction effects for $2{\times}2$ factorial design in randomized complete block design are very superior to powers of parametric statistic without regard to the block size, composition method of effects and the type of population distributions such as exponential, double exponential, normal and uniform. $2{\times}2$ factorial design in RCBD increases error effects and decreases powers of parametric statistic which results in conservativeness. However powers of rank transformed statistic maintain relative preference. In general powers of rank transformed statistic show relative preference over those of parametric statistic with small block size and big effect size.

• Industrial Engineering and Management Systems
• /
• v.6 no.2
• /
• pp.119-124
• /
• 2007

The most difficult and time-intensive issue in the successful implementation of genetic algorithms is to find good parameter setting, one of the most popular subjects of current research in genetic algorithms. In this study, we present a new efficient experimental design method for parameter optimization in a genetic algorithm for general job shop scheduling problem using the Taguchi method. Four genetic parameters including the population size, the crossover rate, the mutation rate, and the stopping condition are treated as design factors. For the performance characteristic, makespan is adopted. The number of jobs, the number of operations required to be processed in each job, and the number of machines are considered as noise factors in generating various job shop environments. A robust design experiment with inner and outer orthogonal arrays is conducted by computer simulation, and the optimal parameter setting is presented which consists of a combination of the level of each design factor. The validity of the optimal parameter setting is investigated by comparing its SN ratios with those obtained by an experiment with full factorial designs.

• Korean Chemical Engineering Research
• /
• v.46 no.4
• /
• pp.756-763
• /
• 2008

The aim of this work was to obtain uniform and well-dispersed spherical silver nanoparticles using statistical design-of-experiment methods. We performed the experiments using 2 k fractional factorial designs with respect to key factors of a general chemical reduction method. The nanoparticles prepared were characterized by SEM, TEM and UV-visible absorbance for particle size, distribution, aggregation and anisotropy. The data obtained were analyzed and optimized using a statistical software, Minitab. The design-of-experiment methods using quantified data enabled us to determine key factors and appreciate interactions between factors. The measured properties of nanoparticles were dominated not only by individual one or two main factors but also by interactions between factors. The appropriate combination of the factors produced small, narrow-distributed and non-aggregated silver nanoparticles of about 30 nm with approximately 10% standard deviation.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.11
• /
• pp.3619-3637
• /
• 2022

Big data analytics offers endless opportunities for operational enhancement by extracting valuable insights from complex voluminous data. Hadoop is a comprehensive technological suite which offers solutions for the large scale storage and computing needs of Big data. The performance of Hadoop is closely tied with its configuration settings which depends on the cluster capacity and the application profile. Since Hadoop has over 190 configuration parameters, tuning them to gain optimal application performance is a daunting challenge. Our approach is to extract a subset of impactful parameters from which the performance enhancing sub-optimal configuration is then narrowed down. This paper presents a statistical model to analyze the significance of the effect of Hadoop parameters on a variety of performance metrics. Our model decomposes the total observed performance variation and ascribes them to the main parameters, their interaction effects and noise factors. The method clearly segregates impactful parameters from the rest. The configuration setting determined by our methodology has reduced the Job completion time by 22%, resource utilization in terms of memory and CPU by 15% and 12% respectively, the number of killed Maps by 50% and Disk spillage by 23%. The proposed technique can be leveraged to ease the configuration tuning task of any Hadoop cluster despite the differences in the underlying infrastructure and the application running on it.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 2004.10a
• /
• pp.700-703
• /
• 2004

A reliability analysis and design procedure based on the design of experiment (DOE) is combined with the response surface method (RSM) for numerical efficiency. The procedure established is based on a 3$^n$ full factorial DOE for numerical quadrature using explicit formula of optimum levels and weights derived for general distributions. The full factorial moment method (FFMM) shows good performance in terms of accuracy and ability to treat non-normally distributed random variables. But, the FFMM becomes very inefficient because the number of function evaluation required increases exponentially as the number of random variables considered increases. To enhance the efficiency, the response surface moment method (RSMM) is proposed. In RSMM, experiments only with high probability are conducted and the rest of data are complemented by a quadratic response surface approximation without mixed terms. The response surface is updated by conducting experiments one by one until the value of failure probability is converged. It is calculated using the Pearson system and the four statistical moments obtained from the experimental data. A measure for checking the relative importance of an experimental point is proposed and named as influence index. During the update of response surface, mixed terms can be added into the formulation.