"SVM and weights"

schillsschills Member Posts: 16  Maven
edited May 23 in Help

Hello

I was wondering if it is possible to apply weights (via "scale by weights") to a set of data before loading it into the SVM to begin learning from the data? Or does the SVM already apply its own weights, thus not allowing my to use any other weights?

If this is possible, how would i go about doing it? I believe i should apply weights after the SVM has learned all the data, but I'm not sure.
I wish to use the SVM to model my data and help me predict the label attribute, but then also use my own set of weights to further enhance the model.

Cheers
Schills
Tagged:

Answers

  • rakirkrakirk Member Posts: 29  Maven
    I read your question two ways:

    How to weight data prior to classification:
    -You could simply add or replace an attribute set with these weights and keep the original label outcome. This is an approach sometimes used, especially with higher-dimensional data, to simplify a set.

    How to apply weights to data after classification:
    -You can run a learning algorithm on the results of the original algorithm. Not always a good idea, but if you know what you want to do and you manage your assumptions, this can be really powerful.
    -You can also use clustering techniques to provide a set of weights after classification.
Sign In or Register to comment.