yes, it's a good idea and from
my perspective easy to do with PHP/MySQL?
And if necessary i want help, but my time-window
for something like this is small, because i'm working in business
instead of university-research.
----- Original Message -----
From: "Marina Santini (ITRI)" <Marina.Santini(a)itri.brighton.ac.uk>
Sent: Friday, October 31, 2003 4:58 PM
Subject: [Wekalist] Request of confirmation: Laplace Estimator and Naive
In the following document:
it is stated:
"The Naïve Bayesian algorithm in Weka is
weka.classifiers.NaiveBayes. By default, it always add 1 to
the number of different values for a particular attribute.
In addition, there is not an option which can be used
to specify the value of µ. In this case,
we need to calculate the result by hand."
I would like to know if this information is true, and
consequently I don't need to modify my attribute values
in order to avoid any zero-probability problems when using the
Naïve Bayes classifier.
This piece of information is not explicitly stated in
the manual ("Data Mining") as far as I know.
Maybe it is worth adding an online search utility to the
the weka archive, able of searching through all the months...
Opening every month and browising through
email subjects/threads can be
time-consuming and inaccurate. This archive is so rich of
information for beginners (like myself) and for experts
that it's a pity it cannot be exploited more thoroughly.
Duplicate questions would decrease considerably
Wekalist mailing list