01-26-2017 05:55 AM
I'm looking at clustering customers based on their transactional data. I have about 3000+ Products (items) in total that belong to various categories (6+) and I need to sort/group the products based on purchasing behavior of the customers. i.e., which products are bought together by the same customer.
I need to use Yule's Q measure of association (between all prooduct combinations) to prepare the input for the cluster analysis. I can pick one category at a time in order to reduce the product cominations.
Yule's Q is (ad - bc)/(ad+bc).
Conceptually, this is the number of pairs in agreement (ad) - the number in disagreement (bc) over the total number of paired observations.
How can I go about doing that using SAS. Proc Distance does not provide the option of Yule's Q
Any help will be more than appreciated.
01-26-2017 07:28 AM
It looks like PROC FREQ can, with Gamma being equivalent to Yule's Q.
It may may help if present some sample data and expected output. Although not all of us understand the specific terminology, most can code many standard formulas.
01-27-2017 06:40 AM