Dear members,
suddenly a question surfaced: how to deal with
severely unbalanced dependent variables in
logistic regression? With 'unbalanced' I mean
cases like 10 dead out of 200 total and so on.
The classification matrix seems biased under
these circumstances. What is the best remedy:
manipulation of the intercept, weighting of the
cases or randomly excluding cases from the larger group?
Sincerely,
Göran Granath
Goran 'Joe' Granath, PhD Phone: +46-18-30 16 67
Ariadne Exploration AB E-mail: [log in to unmask]
Malma Ringväg 1 www.ariadne.se
S-756 45 Uppsala
Sweden
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|