I have watched lectures upto Week3.

In regression, we plot the x and y co-ordinates on a plane and try to draw an imaginary line that reduces the aggregate of spatial distance between the line and the points. This imaginary line is called the boundary. In case of a binary valued output, the plane would be divided into two halves. In case of a real valued output, there can be many boundaries and they can even take different shapes.

Am I right with my conceptual understanding? If I am right, then

- How to visualize in a geometric form if x has lot of features/dimensions? From my limited reading on ML, I have seen that each feature of x could be visualized for y. But is there a more effective way?
- This question may be more ambiguous but is there a way to comprehend the details in a non-geometric form?
- Is the geometric examples given is just a way to understand the problem or is it the only way to solve ML problems?