ABSTRACT

These explanatory variables or covariates influence the lifetime model through the reliability function, and thus by implication through the hazard function. Common to such models is the notion of a baseline reliability function which corresponds to the lifetime behaviour for some standard or initializing condition. The general area of regression modelling covers a range of techniques appropriate to data for which there are covariates. In reliability there is widespread applicability of models based on the Weibull distribution. There are essentially two principal objectives in the modelling process: to establish the form of the 'baseline' model, and to determine how the model parameters vary with the covariates. The experimental procedure giving rise to the data of the above examples ensured that the diameter distribution of the fibres concerned was similar at each length. A mixture model is most distinctive when it leads to a bi- or multimodal distribution.