581994_en-US

liliconcom

Survival analysis: other distributions

A customer asks if for the multivariate survival analysis one can use or select also common models like Gumbo, Gumbel or Gamma besides Weibull and the others mentioned in the documentation (lognormal, exponential, Fréchet, loglogistic, SEV, normal, LEV, and logistic distributions).

Re: Survival analysis: other distributions

I am not sure what the customer means by "multivariate survival analysis". All in JMP now are "univariate survival analysis" in the sense that there is only ONE time-to-event VARIABLE, whose observations may be censored so one will see more than one column of data.

Correspondingly, "multivariate survival analysis" means there are multiple time-to-event variables, which are modeled simultaneously.

The customer could mean survival analysis with multiple explanatory variables.

Now the distributions. "Gumbel" is our LEV.

This is from wiki

This is from JMP Doc

LEV is available in Life Distribution, Fit Life by X, and Parametric Survival.

I am not sure about Gumbo. Could it be a typo?

Gamma is only available in Life Distribution. Gamma is not a commonly used distribution. Though it provides some flexibility compared to Weibull, Lognormal, it has its own set of challenges.

Re: Survival analysis: other distributions

@peng_liu I think what @martindemel was asking is multiple regression for censored data. Believe or not Gamma is available in GenReg :-).

 

Re: Survival analysis: other distributions

Thanks, @jiancao Now it is all covered. And you guys just found something that Clay and I have to trade LOL!

Re: Survival analysis: other distributions

Thanks! this helped a lot. I will check for Gumbo, a short research didn't bring up any useful results, so I'm sure there is a typo on the customers side.