I'm doing a logistic regression with SPSS 17.0.
I've got 10,000 cases.
When I try to check 200 variables (Forward - conditional) - it takes 1
minute (did 40 steps).
When I try 400 variables, it takes 25 minutes (did 140 steps).
Computer: Vista, 3 gb ram, quad-core cpu (the cheapest intel quad core),
plenty of free disk space.
It runs one of the cpu cores to the max, but uses very little RAM (only
100mb).
Why does it slow down so dramatically?
---
Secondary question: I'm analyzing webpage text. So I'm trying to
predict whether a webpage is about a conference. Thus I'm counting how
many times a word occurs on each page (and also looking at two word
combinations), and using those counts as variables. This gives me
3000+ variables to test. Would I be better off using a bayesian algorithm?
Aaron
=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD
Locked
