How do I get the distance between the point and the hyperplane in MATLAB 2021a fitcsvm?

Hi all,

Nested cross-validation method gives me the best model 1x1 ClassificationSVM, please see attached. This models gives an accuracy of 94.53% (using crossval).

I was wondering if there is an easy way to calculate the distance between the point and the hyperplane from this output?

Thanks,

Sahil

Answer 1

[Kshitij Chhabra]

Hi Sahil,

From my understanding you are trying to find the distance of a particular data point from the hyperplane. I can recommmend you using the the "predict" function to get the Classification Score which is the signed distance of a point x form the decision boundary and ranges from -∞ to +∞.

You can learn more about the use of "predict" function here.

Hope this helps!

[Sahil Bajaj]

Hi Kshitij, thanks for your response.

Yes, I used "predict" function to get the classification score. But somehow this classification scores (which represents the signed distance of a point x from the decision boundary) is different from the distance parameter I get using libsvm (as explained here: https://www.mathworks.com/matlabcentral/answers/410858-how-do-i-get-the-distance-between-the-point-and-the-hyperplane-using-libsvm). I am not sure why both "classification score from predict" and "distance" calculation using libsvm as explained in this article are not identical - and if both of these represent different things then which one should I use to get the distance of all data points from the hyperplane?

I would really appreciate any help.

Thanks,

Sahil