Bibliography

(eng): Residual analysis is used commonly in statistical tests of model fit, i.e. of good correspondence between data and model. We shall recall the notion of martingale residuals de¯ned for the case of lifetime models based on failure intensities. Analysis of these residual processes has already been studied by many authors. Nevertheless, Bayes approach to this problem is just developing. We shall present a Bayes procedure of estimation in models containing nonparametric components. Further, Bayes construction of residual processes and model goodness-of-¯t assessing will be proposed and demonstrated on an example with Cox's regression model.