Re: Estimate buying cycle for individual customers with Survival Analysis


On 28-Sep-2005, "alvchua@xxxxxxxxx" <alvchua@xxxxxxxxx> wrote:

> Thanks for your reply, it's fine estimating the most probably time to
> next purchase. I have the purchase history of the customer for a
> variety of computer products. The dataset consists of 2000 small
> business, of which only 10% made 1 or more repeat purchases. The
> variables which I think I can work with are mainly purchase behavior
> (RFM) and product types/characteristics.

There is no way of knowing if you have enough data records and independent
(predictor) variables to enable any sort of accurate prediction. In general,
the more data and predictor variables you have, the better model you can
build.

I am the author of a predictive modeling program called DTREG
(http://www.dtreg.com). If you can send me your data file via e-mail along
with a description of the variables, I'll be happy to run it through DTREG
for you, and we can see if it can generate a predictive model with any
reasonable degree of accuracy.

--
Phil Sherrod
(phil.sherrod 'at' sandh.com)
http://www.dtreg.com (decision tree and SVM predictive modeling)
http://www.nlreg.com (nonlinear regression)
.