Regression with Life Data, Incorrect Minitab Documentation?

C

cmoulder

I am doing a minitab analysis with censored data and one accelerating variable.

I seem to be able to do the analysis just fine, and I can duplicate my model coefficients in reliasoft. However, the equation given in the documention to calculate the failure time for a given percentile does not seem to work.

I've attached the coefficients and results straight from the documentation. Can anyone get the same answers as they have?

Thank you,
Chris
 

Attachments

  • Regression with Life Data, Incorrect Minitab Documentation?
    Minitabregwithlifedata.jpg
    71.3 KB · Views: 235
C

cmoulder

Yes. I'm trying to replicate their results before I use my data. I'm using Minitab 17. The coefficients are the same in Minitab 12, but they switched notation of the shape/scale parameter.

I wouldn't care too much, but I need to calculate a hazard for ~1000 additional samples in another population, and minitab does not let you estimate life for a whole new set of predictor values
 
Last edited by a moderator:

Miner

Forum Moderator
Leader
Admin
See attached file. The issue is with Ep. Despite appearing to be an exponent, it is actually a multiplier. It is also based on the Smallest Extreme Value distribution.
 

Attachments

  • Reliability Regression.xlsx
    31.1 KB · Views: 148
Top Bottom