2.3.6 Elastic Net Model
Elastic network36 is a linear regression model using
L1 and L2 norm as prior regular term training. This combination allows
you to learn a non-zero sparse model with a few parameters, like Lasso,
but it still retains some regular properties like Ridge.The convex
combination of L1 and L2 can be controlled by l1_ratio
parameter.Elastic networks are an iterative method.T he best thing about
elastic networks is that they can always produce efficient
solutions.Because it doesn’t cross paths, the solutions are pretty
good.But the most attractive thing about elastic networks is not their
efficient solution, but their rate of convergence.Elastic networks are
very useful when many features are linked.Lasso is likely to consider
only one of these features at random, while elastic networks prefer
two.In practice, one advantage of the tradeoff between Lasso and Ridge
is that it allows the Ridge stability to be inherited Under rotate.