Dear all,
I would like to characterize the behavior of a certain product over its
lifetime. The purpose is to compare the behavior of experimental
versions of the product vs. a baseline version. It's known that the
behavior of the product follows a roughly trapezoidal pattern: start-up
effect where the performance ramps up, then steady state performance,
finally a degradation at the end of the life. A sensor captures the
performance every two minutes. Is anyone aware of a package that can
be used for such a cradle-to-grave comparison? This is not quite
survival analyses, because I'm interested in more than just 'length of
life'. Interest is in 'length of ramp-up', 'length &
amplitude of
steady-state' and 'length of degradation'. I've considered Time
Series
with seasonality or Fourier analyses to deconstruct the behavior into
periodic component. However, the I'm not aware of TS or Fourier
comparisons for contrasting two populations. Such comparisons are
probably common in image analysis, but that's a foreign field to me.
Thank you,
Paul
Paul Prew
Ecolab R&D
Eagan, MN
CONFIDENTIALITY NOTICE: \ This e-mail communication an...{{dropped:11}}
