Detail publikace

Sample size extension in stratified sampling: Theory and software implementation

Originální název

Sample size extension in stratified sampling: Theory and software implementation

Anglický název

Sample size extension in stratified sampling: Theory and software implementation

Jazyk

en

Originální abstrakt

In this paper, a recently developed method for extension of sample size in Latin Hypercube Sampling (LHS) is presented. The method can be applied when an initial LH-design is employed for the analysis of functions of random vectors. The paper explains how the statistical, sensitivity and reliability analyses of a function can be divided into a hierarchical sequence of simulations with subsets of samples of a random vector in such a way that (i) the favorable properties of LHS are retained (low number of simulations needed for statistically significant estimations of statistical parameters of a function of random vector with low estimation variability); (ii) the simulation process can be halted, e.g., when the estimations reach a certain prescribed statistical significance.An important aspect of the method is that it efficiently simulates subsets of samples of random vectors while focusing on their correlation structure or any other objective function such as some measure of dependence, spatial distribution uniformity, discrepancy, etc. This is achieved by employing a robust algorithm based on combinatorial optimization of mutual ordering of samples. The method should serve preferably as a tool for computationally intensive evaluations of functions where there is a need for pilot numerical studies, preliminary and subsequently refined estimations of statistical parameters, optimization of the progressive learning of neural networks or in design of experiments. The paper presents how the method is implemented into commercial software package FReET.

Anglický abstrakt

In this paper, a recently developed method for extension of sample size in Latin Hypercube Sampling (LHS) is presented. The method can be applied when an initial LH-design is employed for the analysis of functions of random vectors. The paper explains how the statistical, sensitivity and reliability analyses of a function can be divided into a hierarchical sequence of simulations with subsets of samples of a random vector in such a way that (i) the favorable properties of LHS are retained (low number of simulations needed for statistically significant estimations of statistical parameters of a function of random vector with low estimation variability); (ii) the simulation process can be halted, e.g., when the estimations reach a certain prescribed statistical significance.An important aspect of the method is that it efficiently simulates subsets of samples of random vectors while focusing on their correlation structure or any other objective function such as some measure of dependence, spatial distribution uniformity, discrepancy, etc. This is achieved by employing a robust algorithm based on combinatorial optimization of mutual ordering of samples. The method should serve preferably as a tool for computationally intensive evaluations of functions where there is a need for pilot numerical studies, preliminary and subsequently refined estimations of statistical parameters, optimization of the progressive learning of neural networks or in design of experiments. The paper presents how the method is implemented into commercial software package FReET.

BibTex


@inproceedings{BUT107580,
  author="Miroslav {Vořechovský} and Drahomír {Novák} and Radoslav {Rusina}",
  title="Sample size extension in stratified sampling: Theory and software implementation",
  annote="In this paper, a recently developed method for extension of sample size in Latin Hypercube Sampling (LHS) is presented. The method can be applied when an initial LH-design is employed for the analysis of functions of random vectors. The paper explains how the statistical, sensitivity and reliability analyses of a function can be divided into a hierarchical sequence of simulations with subsets of samples of a random vector in such a way that (i) the favorable properties of LHS are retained (low number of simulations needed for statistically significant estimations of statistical parameters of a function of random vector with low estimation variability); (ii) the simulation process can be halted, e.g., when the estimations reach a certain prescribed statistical significance.An important aspect of the method is that it efficiently simulates subsets of samples of random vectors while focusing on their correlation structure or any other objective function such as some measure of dependence, spatial distribution uniformity, discrepancy, etc. This is achieved by employing a robust algorithm based on combinatorial optimization of mutual ordering of samples. The method should serve preferably as a tool for computationally intensive evaluations of functions where there is a need for pilot numerical studies, preliminary and subsequently refined estimations of statistical parameters, optimization of the progressive learning of neural networks or in design of experiments. The paper presents how the method is implemented into commercial software package FReET.",
  booktitle="Safety, Reliability, Risk and Life-Cycle Performance of Structures and Infrastructures",
  chapter="107580",
  howpublished="electronic, physical medium",
  year="2013",
  month="june",
  pages="2907--2914",
  type="conference paper"
}