Lab03: SVR and seismic prediction

[jrasero]

• The outcome variable (Drift) needs to be scaled to get meaningful performance. That, or tune epsilon, but that I think would be more problematic for students to do and to justify without a proper knowledge of the algorithm

• Explicitly mention that formula 1 to 11 in the paper should be explained following both lessons on SVM, so they can see the correspondence between SVM for classification and regression.

• Maybe explicitly say what input features they need to use. They seemed to have a hard time finding this info in the paper.