Weibull distribution using SAS

[sas126]

Hi All,

I have a question on Weibull distribution using SAS.

I know there are some SAS procs that do the work.

I have done warranty analysis on Reliasoft Weibull tool.But now I want to do on SAS.

For a product ,we get the shipment data and the replacement data for specific time period.I want to run Weibull 2p, 3p, Log normal,G Gamma methods and check the distribution.

My goal is to make sure that If the replacement rate is more (product has a problem) and make sure to fix it ASAP.

Any method or idea on how can I proceed on this would be very helpful.

Right now I take shipment column and  (replacements in Nevada format) - load to the realisoft weibull tool and get distribution.

The draw back is that there is lot of manual work (data in to excel , change to nevada format) .Want to get rid of this and make it easier in SAS.

Please advise.

Thanks

Sri

[Rick_SAS]

It sounds like you are doing quality control, so you should look at SAS/QC software: SAS/QC(R) 12.3 User's Guide

In particular, you can fit parametric distributions to data by usiung the CAPABILITY procedure (the UNIVARIATE procedure in Base SAS also has this ability). There is an example of fitting Weibull, lognormal and Gamma in the documentation.

[sas126]

Hi Rick

Thanks for your response.Unfortunately I do not have SAS/QC.

All I have is SAS/STAT and SAS/Graph with EG.

Thanks

Sri

[Rick_SAS]

Then use PROC UNIVARIATE.

[sas126]

Hi Rick

I want to use Proc Lifereg. But before using it , I  will have to right censor my data.Currently I have my data in the stacked format.

Please see the attached image for the data. Can you please suggest me how do i right censor and then use model statement for proc lifereg?

Thanks

Sri

[sas126]

How do I use proc univariate on right censored data?

Example:

Strength Censor

10     0

15     1

16     1

32     0

[Reeza]

What are the rules for censoring?

Use those rules to define a new censor variable in Query tool then your data is in the correct format for proc lifereg.

[sas126]

Thanks for your response.

Yes, But that only works for Weibull 2p ,Lognormal and Gamma.

I want to calculate estimates for 3 Parameter weibull distribution.

Thanks

Sri