1
$\begingroup$

I'm playing with scikit-learn. Looking into the user guide and documentation they say:

A low C makes the decision surface smooth, while a high C aims at classifying all training examples correctly.

The default value is 1.0, which is used by most of the examples. I also found the value C=100.

What is meant by "low" and high for C?

$\endgroup$
2
$\begingroup$

The loss function in SVM consists of both Classification error and Margin error.

Now C is constant that attaches itself to the classification error as given below:

SVM_Error = C*Classfication Error + Margin Error

So, the high value of C means that classification error will be more than the marginal error and therefore model will focus on classifying all the data points accurately. Conversely, if the low value of C means that marginal error will be more than the classification error and therefore model will focus creating the larger margin for the hyperplane

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.