2

I have a simulator which simulates the jobs processed on different machines. I wish to model production of defective parts on the machines. So on a machine, I can model the time before the next defect happens. So should I consider Weibull distribution or Poisson distribution? How to choose? I do not have any data to fit and its up to me to decide the characteristics of the machines.

kosmos
  • 413
  • 4
  • 12
  • 2
    Is this what you meant to ask? The time between defects will be a continuous measurement whereas the Poisson distribution is discrete. – Scortchi - Reinstate Monica Nov 11 '16 at 12:12
  • @Scortchi Yes. But still I am not clear. Arrival times are modeled using Poisson. Why not Weibull? – kosmos Nov 12 '16 at 08:05
  • 2
    If arrival *times* are modelled as having an exponential distribution, then the *counts* of arrivals over a given time interval have a Poisson distribution. See [Relationship between poisson and exponential distribution](http://stats.stackexchange.com/q/2092/17230). – Scortchi - Reinstate Monica Nov 12 '16 at 13:18

0 Answers0