Example of Parametric Regression Survival Fitting
The data table Comptime.jmp contains data on the analysis of computer program execution time whose lognormal distribution depends on the effect Load.
Note: The data in Comptime.jmp comes from Meeker and Escobar (1998, p. 434).
1. Select Help > Sample Data Library and open Reliability/Comptime.jmp.
2. Select Analyze > Reliability and Survival > Fit Parametric Survival.
3. Select ExecTime and click Time to Event.
4. Select Load and click Add.
5. Change the Distribution from Weibull to Lognormal.
6. Click Run.
Figure 14.2 Computing Time Output
When there is only one effect, a plot of the survival quantiles for three survival probabilities are shown as a function of the effect.
Time quantiles are desired for when 90% of jobs are finished under a system load of 5. See Meeker and Escobar (1998, p. 438).
7. Click the Parametric Survival Fit red triangle and select Estimate Quantile.
8. Type 5 in the first row beneath Load.
9. Type 0.9 in the first row beneath p.
10. Click Go.
Figure 14.3 Estimates of Time Quantile
The report estimates that 90% of the jobs will be done by 571 seconds of execution time under a system load of 5.