Which output represent distance from hyperplane in fitcsvm while training data?

While training data on fitcsvm, which output represent distance from the hyoerplane. Please explain.

x= rand([100,2])

y= ones(100,1)

mdl= fitcsvm(x,y)

mdl.BoxConstraints

[~,score]= predict(mdl,x)

Is it correct? Does score term represent the distance of training data from the hyperplane?

Answer 1

[Ayush Modi]

Hi Nainsi,

The score output from "predict" function is the signed distance to the decision boundary. I am assuming by hyperplane, you mean decision boundary, so the answer would be yes.

Please refer to the following MathWorks documentation for more information on the score output of “predict” function:

https://www.mathworks.com/help/stats/classreg.learning.classif.compactclassificationsvm.predict.html#bt74ekh:~:text=%E2%80%94%20Predicted%20class-,scores%20or%20posterior,-probabilities%0Anumeric%20column