Alrighty, I feel incredibly dumb here so please bear with me. I have the following output from minitab from a life regression. Now, what I want to do is in excel write up the model to produce a table of expected survival times for each variable. (Variables A to F are Binary)

However, I can't remember at all what a life regression model looks like. It's been years since I've taken my one GLM course and I didn't realize until now that my textbook titled "An Intro to GLM's" is really just a very basic intro that talks more about seeing if a control and test group have different hazard functions.

Response Variable: Z

Censoring Information Count

Uncensored value 106

Right censored value 206

Censoring value: X= 1

Estimation Method: Maximum Likelihood

Distribution: Weibull

Regression Table

Standard 95.0% Normal CI

Predictor Coef Error Z P Lower Upper

Intercept 5.69727 0.230540 24.71 0.000 5.24542 6.14912

A 2.76352 0.65910 4.19 0.000 1.47163 4.05541

B -1.40814 0.29053 -4.85 0.000 -1.97757 -0.838711

C -1.03633 0.318251 -3.26 0.001 -1.66009 -0.412566

D -1.54800 0.379918 -4.07 0.000 -2.29263 -0.803376

E -1.68173 0.305668 -5.50 0.000 -2.28083 -1.08263

F -0.922788 0.337099 -2.74 0.006 -1.58349 -0.262085

Shape 0.921179 0.0736262

My first instinct in writing my model is:

Failure Time = exp( 5.69727 + 2.76352 *A -1.40814*B -1.03633*C -1.54800*D -1.68173*E -0.922788*F)

But I know I'm not utilizing the shape in the model so something isn't right. Using this seems to be predicting failure times that are way too large. Any help would be wonderful.

However, I can't remember at all what a life regression model looks like. It's been years since I've taken my one GLM course and I didn't realize until now that my textbook titled "An Intro to GLM's" is really just a very basic intro that talks more about seeing if a control and test group have different hazard functions.

Response Variable: Z

Censoring Information Count

Uncensored value 106

Right censored value 206

Censoring value: X= 1

Estimation Method: Maximum Likelihood

Distribution: Weibull

Regression Table

Standard 95.0% Normal CI

Predictor Coef Error Z P Lower Upper

Intercept 5.69727 0.230540 24.71 0.000 5.24542 6.14912

A 2.76352 0.65910 4.19 0.000 1.47163 4.05541

B -1.40814 0.29053 -4.85 0.000 -1.97757 -0.838711

C -1.03633 0.318251 -3.26 0.001 -1.66009 -0.412566

D -1.54800 0.379918 -4.07 0.000 -2.29263 -0.803376

E -1.68173 0.305668 -5.50 0.000 -2.28083 -1.08263

F -0.922788 0.337099 -2.74 0.006 -1.58349 -0.262085

Shape 0.921179 0.0736262

My first instinct in writing my model is:

Failure Time = exp( 5.69727 + 2.76352 *A -1.40814*B -1.03633*C -1.54800*D -1.68173*E -0.922788*F)

But I know I'm not utilizing the shape in the model so something isn't right. Using this seems to be predicting failure times that are way too large. Any help would be wonderful.

Last edited: