A geometric introduction to SVMs
Suppose we have two groups of entities, called B and R, where each entity is described by a pair of numerical coordinates. B includes objects with coordinates (6.38, -10.62), (4.29, -8.99), (8.68, -4.54), and so on and R contains objects with coordinates (6.50, -3.82), (7.39, -3.13), (7.64, -10.02), and so on (the example used in this discussion has been taken from https://scikit-learn.org/stable/modules/svm.html#classification). Plotting these points on a graph gives us the following:
Figure 7.1 – Plot of the R and B points
It looks as though you should be able to draw a straight line to separate the two groups, and if you could, then you could use it to decide whether some new point was an instance of R or B.
There are numerous ways of finding such a line for a simple case like this. One approach would be to find the convex hulls (Graham, 1972) for the two groups – that is, the polygons that include them...