Help using Base SAS procedures

clustering similar behaviour

Reply
Super Contributor
Posts: 395

clustering similar behaviour

Hi,  I am doing a time series (Autoreg) model, which I have over 100 models with 10 years of monthly data.  I have auto prices to over 100 FS's (FS are the 2 digits of a Canadian postal code eg. D2).  I don't want to make 100 models so I need to find what FS's have prices movements in common that I can group them all, but also some cases where it is important to leave some open FS's to more granularity such as FSA's (the first 3 letters of the postal code, eg. D2Y)....so for example leave M5R, M5T, M5H..  hope I've made it simple enough for a good answer.  I basically need a procedure in SAS that allows me to close up some FS (commonality) and leave some open (more detail)..  I was wondering if VARCLUS could work, or any other.   Thanks

sample data

FS       FSA

M5       M5R

M5       M5T

M5       M5H

M4       M4R

M4       M4E

D2        D2K

D2        D2L

D4        D4G

D3        D3F

look something like this

FS       FSA

M5       M5O    <- was M5 and M4     

D2        D2K

D2        D2L

D4        D4O    <- was D4 and D3

Ask a Question
Discussion stats
  • 0 replies
  • 105 views
  • 0 likes
  • 1 in conversation