L1TF trend fitting in python including seasonality and other sparse features. Includes a novel "l1-everything" algorithm which uses l1-norms for everything including the residual.

Example:

test_myl1tf.test_l1tf_on_mock_with_period(period=6,alpha=0.5,eta=1.0)

Based on the paper By Kim, Koh & Boyd

Requires numpy and cvxopt libraries matplotlib required to enable plotting

To test (assuming you have nose installed) run nosetests L1-tf_seasonality/test_myl1tf.py