A Bootstrap Algorithm for Mixture Models and Interval Data in Inter-Comparisons

To combine the information from several laboratories to output a representative value chi(sub r) and its probability distribution function is the main aim of an inter-comparison in Metrology. Here, the proposed procedure identifies a simple model for this probability function, by taking into account only the probability interval estimates as a measure of the uncertainty in each laboratory. A mixture density model is chosen to characterize the stochastic variability of the inter-comparison population considered as a whole. The bootstrap method is applied to approximate the distribution function of the comparison output in an automatic way.
The original document contains color images. All DTIC reproductions will be in black and white. Presented at Algorithms for Approximation IV held in Huddersfield, UK on 16-20 Jul 2001. This article is from ADA412833 Algorithms For Approximation IV. Proceedings of the 2001 International Symposium