"Intelligent" selections for sampling from raw reliability data
Raw reliability data are collected for the estimation of reliability parameters, comparisons of different groups of components, and so on. In existing reliability data books, the samples of taxa where estimation procedures are applied are specified solely on the basis of the opinions of experts; data on components are not used in the definition of the taxa. This report proposes a procedure to define the taxa based on collected data on relevant variables outlined by subject matter experts. The definition of samples is done by intelligent selection on the relevant variables; for each variable, the essential categories where most data lie are selected against those of a few heterogeneous components and consequently infrequent combinations of the essential categories are eliminated. The procedure results a relatively small number of taxa covering most data. The dependence of component variables makes the procedure economical with regard to the exclusion of data and relevant to the identification of errors in the collection or storage of the data.
Bibliographic Reference: Paper presented: Seminar on Reliability Data Analysis and Use, Helsinki (FI), May 17-18, 1995
Availability: Available from (1) as Paper EN 39094 ORA
Record Number: 199511120 / Last updated on: 1995-08-22
Original language: en
Available languages: en