5 ms·
Author here. These are all great points. I chose linear regression because it's simple and easy to understand. There is obviously a simple closed form solution
by mattnedrich 12y ago
Author here. These are all great points. I chose linear regression because it's simple and easy to understand. There is obviously a simple closed form solution to this problem. However, as already mentioned, gradient descent has some nice advantages like scaling. Further, a "close enough" solution/approximation might be sufficient depending on the application.
- grayclhn 12y agoThere are algorithms to add variables and observations to an estimated linear regression model, so scaling is not an advantage of gradient descent. You should mention in the post that there are better ways to estimate this model, even if you're just presenting it as an example.
- nkurz 12y agoOne tiny improvement to the article would be to change the "visualization of the surface" images to to show the part of the surface where the solution is eventually found. The 3D graph cuts off at y = -2, which makes things a little harder to understand.