Prediction

#1
I was wondering what methods you can use if you are trying to predict failure in a product.

I have a product and the following information about the product:

1. The date it was produced
2. number of products produced in the lot
3. The number of products purchased out of that lot
4. months in service before it failed
5. The settlement month (when it was reported broken)

All information comes from warranty data, but I am trying to understand what methods I can use to predict if the failures we see are going to turn into a big issue.

Any assistance would be helpful.

Right now we only have weibull, but we are unsure if we can use weibull if we only have lets say 10 data points.

Thank you in advance.
 

terzi

TS Contributor
#2
Hi gaskinwb,

In order to analyze this get this kind of results you may indeed need to use models. Still, for any of these solutions you will probably need a bigger sample. Maybe you can design an experiment to determine the conditions of the products and use survival analysis to analyze your data, which I think is one of the most common techniques to model failures.