Probability Distribution for Cycle Time
Six Sigma – iSixSigma › Forums › General Forums › New to Lean Six Sigma › Probability Distribution for Cycle Time
 This topic has 6 replies, 5 voices, and was last updated 10 years, 11 months ago by Trish G.

AuthorPosts

October 18, 2010 at 2:21 pm #53617
In dealing with staffing estimates etc I use cycle times for transactions etc. the data is not typically “normal” what is the best distribution to use to predict the expected frequency of transactions that take a certain time?
0October 20, 2010 at 6:50 pm #190896Most timebased data sets are postively skewed due to the fact they are naturally bounded at zero. And you can always consider using the oftendiscussed methods for nonnormality – transformation, subgrouping, nonparametrics, etc. Good luck.
0October 21, 2010 at 12:07 pm #190900
SchwarzParticipant@DaveS002 Include @DaveS002 in your post and this person will
be notified via email.The time between the successive occurance of events is usually modeled by the exponential distribution. This is due to its close relationship to Poisson processes. See any statistics text book for an explanation of how to calculate event probabilities under it.
0October 21, 2010 at 3:13 pm #190903Try Weibull. It’s used for reliability analysis using interval data and it will fit a distribution for you.
0October 21, 2010 at 6:42 pm #190906Thanks , i was also looking at the Lognormal distribution..any thoughts on that one
0October 26, 2010 at 2:44 pm #190912
AndellParticipant@jandell Include @jandell in your post and this person will
be notified via email.First display the data in control chart format. Sometimes “special cause” variation can distort a distribution. If it appears stable, then consider using probability plots with normal and lognormal to start. Weibull can be considered, but lognormal often does the job and it’s easier to work with.
Minitab has a distribution ID utility under its quality and reliability tools. Just be sure to select a subset of distributions, or you’ll get a mountain of outputs.
If you have any questions, give me a yell: [email protected]. I would never dream of charging for a 5minute chat.
0October 26, 2010 at 3:20 pm #190913Great…I will give that a try, thanks much
0 
AuthorPosts
You must be logged in to reply to this topic.