Search code examples
svmlibsvm

Is there any need to scale data in SVM if my weight vector is binary in nature?

My weight vector is SVM training (using Libsvm) will have only feature number and then 1 or 0 i.e representing whether present or NOT present.

it would be like below libsvm format.

1 1:0 2:1 3:0 4:1...9600:1
0 1:1 2:0 3:1 4:0...9600:0
....

As all the measurements are based on one criteria whether present or not? Will scaling be a issue?

Thanks.

Solution

  • No, scaling should not be an issue, because scaling basically does that : scale your data back to a 0-1 range.