Please help me choose the right class

Question

I had a problem choosing the right classifier for my data mining task.

I put web pages using the statistical method and mark them using a scale of 1-4, with 1 being the poorest and 4 the best.

I used to use SVM to train the system, since then I used the binary label (1,0). But now, when I switch to this 4-class label, I need to change the classifier, because I think that the SVM classifier will only work for a classification of two classes (please correct me if I am wrong).

So, could you offer some suggestions here about which classifier is most suitable here for my classification purpose.

Thanks in advance for the suggestions.

Answer 1

There are multi-level SVM. LibSVM has an implementation as well as Weka .

It's usually best to experiment with multiple classifiers to find out which one is best for your data. Choosing the type of classifier and learning algorithm is much less important than choosing a set of functions. You can try naive Bayesian, multi-class SVM, MaxEnt, voted perceptrons or anything your library offers.

Answer 2

" ". SVM ( , libSVM), .

, , . , , .

.

Answer 3

Andrew NG Lecture , ML, , , , , .