Are Weka's SMO and logistic regression classifiers dealing with different
attribute scales ?
For instance, the first attribute range is between 0 and 1, and the second
attribute range is between 0 and 100.
Is the second attribute going to have more weight in the classifiers
decision just because its scale is bigger ?
Or does the classifiers "normalize" all attributes to classifier ?
I have instances with several attributes.
The first attribute range is between 0 and 1, and the second attribute range
is between 0 and 100.
Does the second attribute have more weight in the classifiers decision just
because its scale is bigger, and therefore it varies more ?
Or does the classifiers (such as SMO or logistic regression) do some form of
normalization on the attributes ?
does Weka GUI generate any java code?
if yes where is the location of file?
if no how can we read generated output file in java to see predicted
data,mean values,variance,and so on.
for example if we use gaussian process for regression how can we set
different covariance function and
hyperparameters, and how we can optimize hyperparameters(like gradient
Is there a way of transforming
all that leaves and nodes of a tree builded by WEKA to some kind of
rules that a GIS softwares as ArcGIS (GRASS or SAGA) could followed and
deliver a final map based on the input tree (rules)??
So far I am implementing the WEKA trees manually on ArcGIS, but this procedure is prone of failures and very
I have this output:
=== Classifier model (full training set) ===
M5 pruned model rules
(using smoothed linear models) :
Number of Rules : 2
tiempoEstimado > 33
8.0404 * complejidad=0,3,4,5
+ 8.6842 * complejidad=5
+ 1.0346 * conocimientoLenguaje=0,5,1,3,4
- 11.5163 * eficiencia=5,0,4,3
+ 0.4089 * tiempoEstimado
+ 1.3864 * herramientasSoftware=2,4,5,3
+ 10.4265 * variablesES=3,15,4
+ 1.076 * objetivosRendimiento=0,1,2,5
+ 25.1885 [57/49.529%]
9.0342 * eficiencia=5,0,4,3
+ 9.1282 * herramientasSoftware=2,4,5,3
+ 4.547 * objetivosRendimiento=0,1,2,5
+ 19.3248 [35/75.958%]
but I don't know what is the meaning of the numbers beside each variable and how can I use it them to my implementation.
-Universidad Central "Marta Abreu" de Las Villas. http://www.uclv.edu.cu
-Participe en Universidad 2012, del 13 al 17 de febrero de 2012. Habana.Cuba. http://www.congresouniversidad.cu
-Consulte la enciclopedia colaborativa cubana. http://www.ecured.cu/
I need to text classification, and I am in a situation where I require the
classifier to also give me the probability distribution over the classes,
instead of giving me just one single target class value.
For example, if I have classes A, B and C, I would like the classifier to
say that a test instance belongs to class A with 0.5 probability, to B with
0.3 and to C with 0.2 etc. Ideally, I would like the classifier to also tell
me if an instance belongs to none of the classes (that is, it is an
Is there any classifier in WEKA that can do this?
View this message in context: http://old.nabble.com/Class-membership-probabilities-tp31519876p31519876.ht…
Sent from the WEKA mailing list archive at Nabble.com.