home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.math.stat
- Path: sparky!uunet!portal!ssailer
- From: ssailer@portal.hq.videocart.com (Steve Sailer)
- Subject: Need Help w/Scanner Data Analysis
- Message-ID: <C1F7Es.IA6@portal.hq.videocart.com>
- Organization: VideOcart Inc.
- X-Newsreader: Tin 1.1 PL3
- Date: Mon, 25 Jan 1993 17:32:51 GMT
- Lines: 66
-
- I'd appreciate any thoughts you might have on this stumper in
- marketing statistics:
-
- Information Resources, Inc. (IRI) publishes statistics describing
- the purchasing of tens of thousands of different products sold in
- supermarkets. Among the statistics are "% of households buying."
- This measure cannot be simply aggregated across time. For
- example, if you know that 30% of all households buy cat food in
- one 13 week quarter, you would not assume that over the course of
- a 52 week year, the % of households buying cat food is 4 time 30%
- or 120%. Typically, you would have to pay IRI to retabulate
- purchasing for each household over the desired time period.
- However, we want to find a another way.
-
- What we need is a rough way to quickly estimate % of households
- buying various products during typical 2 week periods, assuming
- we already know from IRI the exact statistics for 52 and 13 weeks
- periods.
-
- The other statistic of relevance that we already have for the 52
- and 13 week periods is "# of purchase occasions among buyers." If
- you multiply % of households buying times number of purchase
- occasions among those buyers, you get the average number of
- purchase occasions among all shoppers. For example, if 30% of the
- shoppers bought cat food, and on average those cat food buyers
- bought it 3.33 times per quarter, then the overall average
- shopper (including both cat lovers and cat haters) bought cat
- food 1.0 times per quarter. For this statistic, in contrast to
- the other two, it _is_ reasonable to extrapolate the average
- annual number of cat food purchases among all shoppers: 1.0 times
- 4 = 4 (assuming no seasonal fluctuations or long term growth or
- decline).
-
- All right, here are some examples to help you think:
-
- 52 Weeks 13 Weeks 2 Weeks
- Cold Cereal - All Brands -------- -------- -------
- % of Households Buying 94.7% 77.0% ?
- Purchase Occasions/Buyer 11.7 3.6 ?
- ----- ---- ----
- Pur Occ/All Shoppers 11.08 2.77 0.43
-
- Kellogg's Corn Flakes
- % of Households Buying 36.7% 17.6% ?
- Purchase Occasions/Buyer 2.9 1.5 ?
- ---- ---- ----
- Pur Occ/All Shoppers 1.06 0.26 0.04
-
- Dry Dog Food - All Brands
- % of Households Buying 31.3% 21.3% ?
- Purchase Occasions/Buyer 6.8 2.5 ?
- ---- ---- ----
- Pur Occ/All Shoppers 2.13 0.53 0.08
-
-
- It has been suggested that a negative binomial distribution is
- the key to the problem, but I'm not qualified to comment on that.
-
- Once again, we greatly appreciate your help with any suggestions
- you might have. Please E-mail or post if you think the question
- is of general interest.
-
-
- --
- Steve Sailer
- VideOcart, Inc. ssailer@hq.videocart.com
-