Re: fitness treshold

Ok, i try.
My problem is classificate some example of training in three classes with
(one or more) hypotesis most performant. I have thought to count how many
positive examples and how many negative examples there are. For example an
hypotes of this type "1, ?, 1, 2, 1, ... , 3" has 180 positive examples and
36 negative examples. ? means don't care, anithing value. 1- 2 -3 are
possible values for the nine attributes. 3, the last value, is the value of
target concept. (the target concept, by the way, is the contracceptive
method used by a woman)
Is that correct?
For the treshold fitness, i've thought to define a percent (33% ?), because,
in my example, i have: 180 + and 36 -, so i have a percent of 180/216. It's
right?

thanks
Alessia


"Wesam ELSHAMY" <wesamelshamy@xxxxxxxxx> ha scritto nel messaggio
news:1170846399.948604.61880@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
On Feb 6, 4:44 pm, "Alessia" <alic...@xxxxxxxxx> wrote:
Hello to everybody,

i have some problem with the definition of the fitness treshold. In
particular, i have a problem of this type: nine attributes plus the
target
attribute with three values (1 - 2 - 3). The training examples are in
total
1473 (about 500 for each class target). In this case, which is the
treshold?
Does depend by the number of class (so 33%)?

Thank you a lot!
Alessia

Your question isn't clear (at least for me).

You may rephrase the question and explain the context of the problem
you are facing.



.