In the formulation in Exercise 10.16, the implicit assumption made was that the problem was separable by shift. Why might this not be the case in practice? What other factors might affect the formulation?

Data From Exercise 10.16:

Consider the publicly available heart disease data set at (http://

archive.ics.uci.edu/ml/datasets/heart+Disease). Use the 13 attributes given to predict angiographic disease (binary classification)

using an SVM approach. Train on half of the data, and validate on the second half.