Warranty analysis with field data

[Steve Larson]

I am trying to do a warranty analysis to look at field reliability and projected warranty exposure. For the products we are interested in, all we have is monthly shipment and warranty return data. We do not have the return data segmented by lot or shipment month - just the total returns for the outstanding installed base at that time.

The Wiebull++ warranty tool would handle this IF I had the return data segmented by shipment month, but I do not. Do I have to assume some distribution of the monthly return data by shipment month?

[Pantelis]

Since age data is impossible to ascertain, trying to fit a Weibull or another distribution to this data is not going to work without making some assumptions. Depending on the assumptions made different analyses can be utilized.

In general for this type of data a non-parametric approach may be better.