Cycle Time Data – Mean or Median?
Six Sigma – iSixSigma › Forums › Old Forums › Finance › Cycle Time Data – Mean or Median?
- This topic has 3 replies, 4 voices, and was last updated 17 years, 5 months ago by
Menco van der Weerd.
-
AuthorPosts
-
February 7, 2005 at 4:08 pm #24321
Hi – I’m in the process of collecting cycle time data of a loan closing process. One of the problems is that the data is skewed to the right because some loans are put on hold at the request of the borrower (or for many other outside factors) significantly delaying the closing date. Those influences are not the focus of this project – the scope is limited to our internal processing system. And, unfortunately, it’s very difficult to determine in a reliable way which loans were delayed due to outside causes.
Should I be using the mean or median or some other measure as the measure of central tendency?0February 7, 2005 at 4:20 pm #58346From your post it seems the issue is with the Variation in the Cycle Time for Loan Processing then why dont you look at SPAN and try and reduce the same. For Central Tendency even if you had removed the data for causes like on hold due to client request etc, will the data be normal?? I am guessing it might not be, so go for Median.Any other ideas on this are most welcome
0February 10, 2005 at 11:26 am #58347
Tajjammul HussainMember@Tajjammul-HussainInclude @Tajjammul-Hussain in your post and this person will
be notified via email.I agree, because the the nature of cycle time is such that in majority of cases the data will be skewed and in that case median makes sense along with a focus variation.
0February 23, 2005 at 9:32 am #58367
Menco van der WeerdParticipant@Menco-van-der-WeerdInclude @Menco-van-der-Weerd in your post and this person will
be notified via email.The median is the way to go in skewed cycletime data. Also the spread between the Q3 and Q2 (Q2 equals median) could be interesting. If the difference is large then you’re likely to have multiple influence factors maybe even multiple populations. If the difference is small the underlying problem is likely caused by outliers. Same goes for the spread between Q3 and the maximum.Hunting for the outliers and the underlying reasons for the outliers is in my experience a good way to bring the process under control, although it probably never will conform to a normal distribution. This can be done by a Best-of-the-Best vs Worst-of-the-worst analyses. Often this will result in a split of the original process in multiple processes, in line with the cause for the outliers or the influencefactors.
RegardsMenco van der Weerd
0 -
AuthorPosts
The forum ‘Finance’ is closed to new topics and replies.