"Intelligent" selections for the statistical analysis of reliability database dataFunded under: JRC-REACTSAFE 5C
Reliability databases store data on components and component related events in a large number of variables, occasionally with a large number of categories per variable. This great detail poses difficulties in the definition of samples to be drawn from databases for purposes of estimation and testing. However, database variables are highly dependent on each other and the categories taken up by the variables are not close to being equi-distributed. Hence, samples from databases may be drawn by so-called Intelligent Selections: a few codes or combinations of codes where most data lie are selected and the remaining codes - where a few heterogeneous components lie - are not considered in the subsequent statistical analysis. Intelligent selections on component characteristics of electric motors in the Component Event Data Bank are applied to draw samples for estimation of and testing on reliability parameters.
Bibliographic Reference: Paper presented: European Safety and Reliability Conference, RisÝ National Laboratory (DK), June 10-12, 1992
Availability: Available from (1) as Paper EN 36431 ORA
Record Number: 199210974 / Last updated on: 1994-12-02
Original language: en
Available languages: en