copt.LogisticLoss

class copt.LogisticLoss(A, b, alpha=0, intercept=True)

Logistic regression loss function with L2 regularization

This loss function is very popular for binary classification tasks. Labels (b) are assumed to be 1 or -1.

References

Loss function and gradients for full gradient methods are computed as detailed in to http://fa.bianp.net/blog/2013/numerical-optimizers-for-logistic-regression/

__init__(A, b, alpha=0, intercept=True)

Methods

__init__(A, b[, alpha, intercept])
gradient(x)
lipschitz_constant([kind])
partial_gradient_factory()