Support Vector Machine (SVM) separates two classes by drawing a line that has maximize distance with the closest points. The plane being drawn is called hyperplane. SVM works with two steps:
- Make lines that separate the two classes.
- Choose the most ideal line which indicated by the highest distance to the nearest points.
In SVM, there is a parameter called C that is used to adjust the trade off between: the wide of the margin and the accuracy of the classification. The graphs below shows how the change in different value of C.
In more simply way, we can say that C is parameter for controlling error.
What about Gamma?
Gamma defines how far the influence of a single training example reaches. Low values -> far , while high value -> close.
Program Link 