Data set requisites:
1. Must be discrete-time;
2. Should provide the measured inputs (“Xt”, the controllable factor values)
and output values (“Yt”, the values of the observed quality characteristic
thought to be affected by Xt);
3. Should be at least 75 time units long, where the time unit is defined by
the context of the data.
4. The data must not have been obtained while a (linear) feedback controller
was operating on the process, i.e., this must be “open loop” data. To check
whether you have met this requisite, use Minitab to determine the
cross-correlation function (CCF) between the input series and the output
series. If the CCF at lag zero, lag one, or lag two is significant, you
probably have linear feedback.
5. I prefer data sets for which there is a “story” related about what the
process is, about what are the inputs and outputs, about how they were
measured, etc. Please include this information in your report.
6. Be sure you provide time series data from a dynamic process; not data
from a responsive process.