Life Distribution with defective subpopulation

martindemel

Hi,

Attached there is an data set from a customer with defective subpopulations. He uses life distribution, and wants to fit a DS Weibull distribution as Weibull does not a good job because of defective subpopulation.

The issue is that with the original data DS Weibull shows a fit, but no confidence interval and the message says “Failure: Cannot decrease objective function” and “Convergence failed: Cannot decrease objective function”. After creating a new column with a log10 transformation of the time data I am able to fit a DS Weibull.

Can someone explain to me why this is the case (both I cannot see the confidence intervals and why there is a difference if log transformed or not)?

The customer is Robert Bosch and the data is confidential. But Bosch could become a good opportunity in case the evaluation works out.

Thanks,

Martin

SEnden.jmp

malcolm_moore1

Try excluding and hiding row 1 Martin which seems to be a censored observation at time zero. DS Weibull seems to fit ok if you exclude this point and I'm not sure of the logic of a censored observation at time zero!