Predictive modeling

[Lopa2016]

Hi,

I have the following sample dataset

ID	Purchase Date	Age	Sales
3621901	04/01/2010	21	990
6271017	04/01/2010	29	642
3914676	02/01/2010	26	955
3888167	04/01/2010	32	834
3888167	04/01/2010	30	616
3888167	04/01/2010	38	657
3289156	04/01/2010	33	322
3621901	30/11/2011	27	855
6271017	31/05/2012	40	381
3914676	11/06/2012	29	915
3888167	11/06/2012	40	972
3015141	28/05/2012	26	755
4518247	28/05/2012	31	415
4518247	13/06/2012	30	652

having 3 years of data for different customers & as seen the customers ids may repeat over 2 different point in time.How is it best possible to run a predictive model given there are different point of time involved with same customer ids?

Would be great if you can advice.

[Ksharp]

It seems it is REPEATED measure model.

Check Generalize Linear Mixed Model.

PROC GLIMMIX + RANDOM year/subject=id residual ;