Date: Thu, 1 Jun 2006 11:42:05 -0400
Reply-To: Peter Flom <Flom@NDRI.ORG>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Peter Flom <Flom@NDRI.ORG>
Subject: Re: Strange Statistic Problem
In-Reply-To: <E68A1768216E394593B3A390F77725A407B5BD2A@MI8NYCMAIL06.Mi8.com>
Content-Type: text/plain; charset=US-ASCII
>>> Rajat Mathur <RMathur@INDUCTIS.COM> 6/1/2006 9:18 am >>>
I am analyzing a call center data and I have to figure out frequent and
infrequent callers based upon the number of calls they have made in past one year. This is the format of information available to me:
No of Calls
Percentage of Customer
1 40%
2 20%
3 10%
4 6%
5 5%
12 0.8%
Greater then 12 ....
My question is what is the best statistical way to find out a cutoff of
calls which can differentiate between frequent callers and infrequent
callers?
Which means can I statistically say that callers who have called more
then 5 times in an year are frequent callers?
>>>
This is not a statistics question, and there is, AFAIK, no statistical answer to it. 'Frequent' is not statistically definable. If someone calls once a month, is that frequent? Well, I call SAS tech support about that often, and I guess I am a pretty frequent caller, but not amazingly so. If I called the support for my refrigerator once a month, I would regard that as amazingly frequent. It depends on the substantive area.
One could, if one had to, try to find a distribution which fit the data reasonably well (I suggest starting with Poisson and Negative Binomial) and then try to find outliers, but that does not answer the question you posed.
HTH
Peter
Peter L. Flom, PhD
Assistant Director, Statistics and Data Analysis Core
Center for Drug Use and HIV Research
National Development and Research Institutes
71 W. 23rd St
http://cduhr.ndri.org
www.peterflom.com
New York, NY 10010
(212) 845-4485 (voice)
(917) 438-0894 (fax)
